Skip to main content
Technology Collaboration Programme by IEA

Expert elicitation · results

What 104 experts expect for hydrogen-based iron and steelmaking

Pooled judgements of the Task 52 expert network on technology readiness, supply chains, finance, pathways, regional routes and policy, with the spread of views shown as well as the midpoint.

Experts
104
Countries
20
Modules
9

Key findings

Real but slow, and held back by markets more than by technology

Between 12 September and 8 November 2026, 104 experts from the IEA Hydrogen TCP Task 52 network completed a structured online survey on hydrogen-based iron and steelmaking. They come from 20 countries and from industry, universities and research institutes, governments, intergovernmental bodies and NGOs. Each expert answered the modules they know best. Quantities were asked as a low, best and high estimate, and dates as probabilities, so uncertainty is also recorded. The result is not another model scenario. It is a measured record of what the people building, financing, regulating and studying this industry expect, and of where they disagree.

The survey covers the whole chain in nine linked modules: ore quality, hydrogen supply, plant flexibility, trade routes, power systems, finance, policy and buyer behaviour. It reports the spread of opinion as well as the central view. Experts agree that scrap-based EAF is the safe investment and that green iron will be made where firmed renewable power is cheapest. They disagree on how much iron will be traded, and when. The survey also asked what knowledge is missing. Carbon accounting and certification rules top that list, which tells Task 52 and its partners where work is needed most.

Their message is clear: hydrogen-based direct reduction is coming, but slowly. They expect it to be commercially proven at shaft-furnace scale by the end of this decade, and to make about 6% of the world's steel by 2040. Costs converge with the blast furnace in the late 2030s. What holds it back is not the technology. It is offtake, energy prices and whether policy support will last. Four in five experts name uncertain offtake and the green premium as the main barrier to investment.

The twelve findings below summarise the results. Each links to the full result and its chart.

  1. 1

    Hydrogen DRI makes 6% of steel in 2040

    We asked the experts Allocate global crude steel production in 2040 across production routes (percent, summing to 100). Q8

    0%20%40%60%80%100% BF-BOF, unabated: median 42.0%, IQR 30.0–50.0, range 5.0–71.0 (n=88) BF-BOF, unabated Reference: 70% 42 BF-BOF with CCUS and/or hydrogen injection: median 5.0%, IQR 2.8–10.0, range 0.0–32.0 (n=88) BF-BOF with CCUS and/or hydrogen in… Reference: 0% 5 Natural-gas DRI-EAF, with or without CCS: median 12.0%, IQR 10.0–15.0, range 4.0–30.0 (n=87) Natural-gas DRI-EAF, with or withou… Reference: 7% 12 Hydrogen-based DRI-EAF, including imported green iron: median 5.6%, IQR 3.0–10.0, range 0.0–30.0 (n=88) Hydrogen-based DRI-EAF, including i… Reference: 0% 6 Scrap-based EAF: median 30.0%, IQR 25.0–35.0, range 10.0–60.0 (n=88) Scrap-based EAF Reference: 22% 30 Other novel routes (electrolysis, smelting reduction, plasma): median 2.0%, IQR 1.0–3.0, range 0.0–10.0 (n=79) Other novel routes (electrolysis, s… Reference: 1% 2

    Box: interquartile range · thick line: median · whiskers: min–max · number: median · purple line: reference value

    Hydrogen-based DRI-EAF is expected to make 6% of global crude steel in 2040 (median; interquartile range 3 to 10%), behind unabated BF-BOF at 42%, scrap-based EAF at 30% and natural-gas DRI-EAF at 12% Q8, n=88.

  2. 2

    Ten million tonnes of capacity around 2037

    We asked the experts In which year will at least 10 million tonnes per year of hydrogen-based DRI capacity be in commercial operation globally? Q9

    10% chance by

    2032

    IQR 2030–2035 · never: 1

    50% chance by

    2037

    IQR 2035–2040 · never: 1

    90% chance by

    2042

    IQR 2040–2050 · never: 2

    0%25%50%75%100%20272031203520392043204720502027: 3% pooled probability2028: 4% pooled probability2029: 6% pooled probability2030: 10% pooled probability2031: 15% pooled probability2032: 20% pooled probability2033: 26% pooled probability2034: 32% pooled probability2035: 38% pooled probability2036: 45% pooled probability2037: 50% pooled probability2038: 56% pooled probability2039: 61% pooled probability2040: 67% pooled probability2041: 70% pooled probability2042: 74% pooled probability2043: 77% pooled probability2044: 79% pooled probability2045: 82% pooled probability2046: 84% pooled probability2047: 86% pooled probability2048: 87% pooled probability2049: 89% pooled probability2050: 90% pooled probability

    Pooled cumulative probability: average of each expert's piecewise-linear curve through their 10%, 50% and 90% years.

    Ten million tonnes a year of hydrogen-based DRI capacity is expected in the late 2030s: the pooled probability curve crosses 50% in 2037, and the median expert puts the even-odds year at 2037 (IQR 2035 to 2040) Q9, n=89.

  3. 3

    Cost parity only in the late 2030s

    We asked the experts What production-cost premium will hydrogen-based DRI-EAF crude steel carry over BF-BOF in the best-in-class location, excluding carbon pricing and subsidies? Q10

    2030 2030 · Below BF-BOF cost: 1% (n=84)2030 · 0 to 10%: 6% (n=84)2030 · 10 to 25%: 19% (n=84)2030 · 25 to 50%: 46% (n=84)2030 · Above 50%: 27% (n=84) Mean score (n=84)37.4 2035 2035 · Below BF-BOF cost: 1% (n=83)2035 · 0 to 10%: 13% (n=83)2035 · 10 to 25%: 42% (n=83)2035 · 25 to 50%: 35% (n=83)2035 · Above 50%: 8% (n=83) Mean score (n=83)26.1 2040 2040 · Below BF-BOF cost: 5% (n=82)2040 · 0 to 10%: 37% (n=82)2040 · 10 to 25%: 40% (n=82)2040 · 25 to 50%: 13% (n=82)2040 · Above 50%: 5% (n=82) Mean score (n=82)16.6
    Below BF-BOF cost0 to 10%10 to 25%25 to 50%Above 50%Number at right: mean score

    In 2030, 74% of experts expect hydrogen DRI-EAF steel to cost at least 25% more than BF-BOF steel in the best location, before carbon prices and subsidies. By 2040, 82% expect the premium to be 25% or less, and 41% expect 10% or less Q10, n=84.

  4. 4

    Offtake, not technology, blocks investment

    We asked the experts Rank the three biggest barriers to reaching final investment decision on commercial-scale hydrogen-based ironmaking projects today. Q11

    Offtake and green-premium uncertainty: 63% · 1st 44% Offtake and green-premium uncertainty 63% · 1st 44% Hydrogen or electricity price risk: 53% · 1st 32% Hydrogen or electricity price risk 53% · 1st 32% Durability and predictability of policy support: 20% · 1st 4% Durability and predictability of policy suppo… 20% · 1st 4% Carbon price and CBAM implementation uncertainty: 18% · 1st 7% Carbon price and CBAM implementation uncertai… 18% · 1st 7% Cost of capital and balance-sheet constraints: 13% · 1st 4% Cost of capital and balance-sheet constraints 13% · 1st 4% First-of-a-kind technology risk: 9% · 1st 3% First-of-a-kind technology risk 9% · 1st 3% Capex escalation and EPC capacity: 8% · 1st 3% Capex escalation and EPC capacity 8% · 1st 3% Permitting and grid-connection delays: 7% · 1st 1% Permitting and grid-connection delays 7% · 1st 1% Security of DR-grade ore supply: 7% · 1st 1% Security of DR-grade ore supply 7% · 1st 1% Skilled workforce availability: 2% · 1st 1% Skilled workforce availability 2% · 1st 1%

    Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

    The two barriers to final investment decision that dominate everything else are offtake and green-premium uncertainty (ranked first by 44%, mentioned by 80%) and hydrogen or electricity price risk (ranked first by 32%, mentioned by 70%). First-of-a-kind technology risk is mentioned by only 13% Q11, n=98.

  5. 5

    Lead markets and grants now, carbon pricing later

    We asked the experts Rank the three policy instruments that would most accelerate hydrogen-based ironmaking deployment over the next five years. Q12

    Lead markets or quotas for near-zero steel: 33% · 1st 16% Lead markets or quotas for near-zero steel 33% · 1st 16% Capital grants for first plants: 28% · 1st 17% Capital grants for first plants 28% · 1st 17% Carbon contracts for difference or production subsidies: 25% · 1st 14% Carbon contracts for difference or production… 25% · 1st 14% Domestic carbon pricing: 23% · 1st 17% Domestic carbon pricing 23% · 1st 17% Hydrogen production tax credits or subsidies: 21% · 1st 9% Hydrogen production tax credits or subsidies 21% · 1st 9% Green public procurement: 17% · 1st 4% Green public procurement 17% · 1st 4% Border carbon adjustment (CBAM): 16% · 1st 8% Border carbon adjustment (CBAM) 16% · 1st 8% Bilateral green iron and steel trade agreements: 16% · 1st 7% Bilateral green iron and steel trade agreemen… 16% · 1st 7% Product standards and embodied-carbon limits: 8% · 1st 3% Product standards and embodied-carbon limits 8% · 1st 3% Concessional finance, export credit or blended finance: 8% · 1st 3% Concessional finance, export credit or blende… 8% · 1st 3% Harmonised definitions and certification: 4% · 1st 1% Harmonised definitions and certification 4% · 1st 1%

    Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

    For the next five years, experts favour lead markets or quotas for near-zero steel and capital grants for first plants, now almost level, with carbon contracts for difference third Q12, n=92. After 2030 they trust domestic carbon pricing (mean 4.1 of 5) and border carbon adjustment (also 4.1) most to sustain the industry Q66, n=33, and a rising, predictable carbon price is the clearest least-regret policy, ranked first by 56% Q75, n=36.

  6. 6

    Gas first, hydrogen later

    We asked the experts For commercial-scale DRI projects reaching final investment decision in 2026 to 2028, which initial reductant configuration do you expect to be most common? Q16

    64% expect new DRI projects to start on natural gas and switch to hydrogen later

    Gas first, hydrogen later: 64% expect commercial DRI projects reaching investment decision in 2026 to 2028 to start on natural gas with a planned transition to hydrogen Q16, n=92, and 65% expect natural gas or methane injection to remain the dominant source of carbon in hydrogen DRI in 2035 Q26, n=43.

  7. 7

    Hydrogen shaft furnaces are nearly ready

    We asked the experts When will the first commercial-scale plant (at least 1 Mt/yr) be operating for each route? Q18

    Shaft-furnace DRI on 100% hydrogen Shaft-furnace DRI on 100% hydrogen · Already operating: 8% (n=39)Shaft-furnace DRI on 100% hydrogen · By 2028: 41% (n=39)Shaft-furnace DRI on 100% hydrogen · 2029 to 2032: 41% (n=39)Shaft-furnace DRI on 100% hydrogen · 2033 to 2036: 3% (n=39)Shaft-furnace DRI on 100% hydrogen · 2037 to 2040: 5% (n=39)Shaft-furnace DRI on 100% hydrogen · After 2040: 3% (n=39) Mean score (n=39)2029.5 Fluidised-bed hydrogen reduction of fin… Fluidised-bed hydrogen reduction of fines · Already operating: 6% (n=36)Fluidised-bed hydrogen reduction of fines · 2029 to 2032: 22% (n=36)Fluidised-bed hydrogen reduction of fines · 2033 to 2036: 56% (n=36)Fluidised-bed hydrogen reduction of fines · 2037 to 2040: 8% (n=36)Fluidised-bed hydrogen reduction of fines · After 2040: 8% (n=36) Mean score (n=36)2034.3 Hydrogen DRI plus electric smelting fur… Hydrogen DRI plus electric smelting furnace · Already operating: 3% (n=37)Hydrogen DRI plus electric smelting furnace · By 2028: 8% (n=37)Hydrogen DRI plus electric smelting furnace · 2029 to 2032: 43% (n=37)Hydrogen DRI plus electric smelting furnace · 2033 to 2036: 27% (n=37)Hydrogen DRI plus electric smelting furnace · 2037 to 2040: 16% (n=37)Hydrogen DRI plus electric smelting furnace · After 2040: 3% (n=37) Mean score (n=37)2032.8 Hydrogen plasma or smelting reduction Hydrogen plasma or smelting reduction · 2029 to 2032: 3% (n=35)Hydrogen plasma or smelting reduction · 2033 to 2036: 20% (n=35)Hydrogen plasma or smelting reduction · 2037 to 2040: 34% (n=35)Hydrogen plasma or smelting reduction · After 2040: 40% (n=35)Hydrogen plasma or smelting reduction · Never: 3% (n=35) Mean score (n=35)2040.1 Molten oxide electrolysis Molten oxide electrolysis · Already operating: 3% (n=35)Molten oxide electrolysis · By 2028: 3% (n=35)Molten oxide electrolysis · 2033 to 2036: 14% (n=35)Molten oxide electrolysis · 2037 to 2040: 20% (n=35)Molten oxide electrolysis · After 2040: 46% (n=35)Molten oxide electrolysis · Never: 14% (n=35) Mean score (n=35)2040.5 Low-temperature iron electrowinning Low-temperature iron electrowinning · 2029 to 2032: 3% (n=35)Low-temperature iron electrowinning · 2033 to 2036: 26% (n=35)Low-temperature iron electrowinning · 2037 to 2040: 17% (n=35)Low-temperature iron electrowinning · After 2040: 49% (n=35)Low-temperature iron electrowinning · Never: 6% (n=35) Mean score (n=35)2040.5
    Already operatingBy 20282029 to 20322033 to 20362037 to 2040After 2040NeverNumber at right: mean score

    Shaft-furnace DRI on 100% hydrogen is rated TRL 7 to 8 by 71% of the technology module Q17, n=41, but the first commercial-scale plant of at least 1 Mt a year has slipped in the network's eyes: 49% expect one to be operating by 2028 or already, and a further 41% say 2029 to 2032 Q18, n=39. Hydrogen supply variability and buffering is the leading technical barrier to scale, ranked first by 63% Q21, n=41.

  8. 8

    4 MWh of electricity per tonne of steel

    We asked the experts What total electricity input per tonne of crude steel will a best-in-class hydrogen-based DRI-EAF plant need in 2035, including electrolysis but excluding mining and pelletising? Q20

    4.0 MWh per tonne of crude steel for a best-in-class hydrogen DRI-EAF plant in 2035

    A best-in-class hydrogen DRI-EAF plant in 2035 is expected to need 4.0 MWh of electricity per tonne of crude steel (IQR 3.5 to 4.5; Q20, n=33), and a first-of-a-kind plant is expected to run at 65% of nameplate over its first two years (IQR 55 to 70%; Q25, n=33).

  9. 9

    USD 165 per tonne of CO2 for parity in Europe

    We asked the experts What carbon price would make hydrogen-based DRI-EAF steel cost-competitive with unabated BF-BOF in Europe in 2030, without other subsidies? Q61

    USD 165 carbon price per tonne of CO2 for hydrogen steel to match unabated BF-BOF in Europe in 2030

    Delivered low-emission hydrogen in a best-in-class location is expected to cost USD 3.0 per kilogram in 2035 (IQR 2.5 to 4.4; Q62, n=34). A carbon price of USD 165 per tonne of CO2 (IQR 141 to 200) would make hydrogen DRI-EAF steel competitive with unabated BF-BOF in Europe in 2030 without other subsidies, higher than the USD 155 the first 84 responses implied Q61, n=30, and the pooled curve puts even odds of European cost parity, with carbon pricing and CBAM but no subsidies, in 2040 Q68, n=27.

  10. 10

    Trade is the least settled topic

    We asked the experts What share of global hydrogen-based DRI/HBI output will be traded internationally as merchant iron in 2040, rather than consumed at the site of production? Q13

    30% of hydrogen-based iron traded internationally in 2040; a liquid HBI market around 2040

    Trade is the least settled topic. Experts expect 30% of hydrogen-based DRI/HBI output to be traded internationally in 2040 (IQR 20 to 50%; Q13, n=71) and 35 Mt a year of merchant hydrogen-based iron to be traded in 2035 (IQR 12.5 to 60; Q44, n=24); both are among the widest spreads in the whole elicitation. A liquid merchant HBI market is expected around 2040 Q53, n=22.

  11. 11

    Buyers need a shared rulebook

    We asked the experts How much does each knowledge gap limit decisions by industry and governments today? Q85

    3.9 of 5 how much carbon accounting and certification rules limit decisions today, the top-rated knowledge gap

    The knowledge gap that most limits decisions today is carbon accounting and certification rules (mean 3.9 of 5; Q85, n=63), ahead of comparisons of national policy packages and power-system integration. Buyers are expected to pay USD 50 to 100 per tonne for near-zero steel in 2030 (35%; Q83, n=66), and automotive is the only sector most experts see making rapid progress or largely transitioning by 2035 (56%; Q84, n=68).

  12. 12

    Scrap-based EAF is the safe bet

    We asked the experts How robust is each pathway if renewable hydrogen still costs more than USD 4 per kilogram in 2035? Q74

    Hydrogen DRI at the existing steel site Hydrogen DRI at the existing steel site · Robust: still the right choice: 7% (n=28)Hydrogen DRI at the existing steel site · Tolerable: viable with support: 46% (n=28)Hydrogen DRI at the existing steel site · At risk of stranding: 46% (n=28) Mean score (n=28)0.6 Importing green HBI from ore-and-renewa… Importing green HBI from ore-and-renewables hubs · Robust: still the right choice: 26% (n=27)Importing green HBI from ore-and-renewables hubs · Tolerable: viable with support: 63% (n=27)Importing green HBI from ore-and-renewables hubs · At risk of stranding: 11% (n=27) Mean score (n=27)1.1 Natural-gas DRI with CCS Natural-gas DRI with CCS · Robust: still the right choice: 35% (n=26)Natural-gas DRI with CCS · Tolerable: viable with support: 38% (n=26)Natural-gas DRI with CCS · At risk of stranding: 27% (n=26) Mean score (n=26)1.1 Blast furnace with CCUS Blast furnace with CCUS · Robust: still the right choice: 21% (n=24)Blast furnace with CCUS · Tolerable: viable with support: 21% (n=24)Blast furnace with CCUS · At risk of stranding: 58% (n=24) Mean score (n=24)0.6 Scrap-based EAF expansion Scrap-based EAF expansion · Robust: still the right choice: 85% (n=27)Scrap-based EAF expansion · Tolerable: viable with support: 15% (n=27) Mean score (n=27)1.9 Electric smelter on lower-grade ore wit… Electric smelter on lower-grade ore with hydrogen DRI · Robust: still the right choice: 16% (n=25)Electric smelter on lower-grade ore with hydrogen DRI · Tolerable: viable with support: 48% (n=25)Electric smelter on lower-grade ore with hydrogen DRI · At risk of stranding: 36% (n=25) Mean score (n=25)0.8
    Robust: still the right choiceTolerable: viable with supportAt risk of strandingNumber at right: mean score

    Scrap-based EAF capacity and scrap sourcing is the clearest least-regret investment, ranked first by 61% Q72, n=36, and 85% judge it robust even if renewable hydrogen still costs more than USD 4 per kilogram in 2035; 58% judge blast furnaces with CCUS at risk of stranding in that world Q74, n=28.

The expert base

Who responded

One response per expert: four in ten from universities and research institutes, the rest from industry, policy, intergovernmental bodies and NGOs, with Europe, a global focus, Australia and China the most-named regions of expertise.

Task 52 affiliation n=103
Subtask A — Technology assessment and readiness: 26% (27) Subtask A — Technology assessment and readine… 26% (27) Subtask B — Resources, supply chains and infrastructure: 16% (16) Subtask B — Resources, supply chains and infr… 16% (16) Subtask C — Global pathways, competitiveness and power systems: 18% (19) Subtask C — Global pathways, competitiveness … 18% (19) Task linkage and exchange — contributing across more than one Subtask, policy advice and international exchange: 12% (12) Task linkage and exchange — contributing acro… 12% (12) Industry or policy reference expert: 20% (21) Industry or policy reference expert 20% (21) Other / not yet assigned: 8% (8) Other / not yet assigned 8% (8)
Sector n=103
Industry — iron or steel producer: 8% (8) Industry — iron or steel producer 8% (8) Industry — mining or ore supplier: 4% (4) Industry — mining or ore supplier 4% (4) Industry — technology provider, EPC or equipment: 10% (10) Industry — technology provider, EPC or equipm… 10% (10) Industry — energy, hydrogen or utility: 4% (4) Industry — energy, hydrogen or utility 4% (4) Finance or investment: 0% (0) Finance or investment 0% (0) Government or public agency: 4% (4) Government or public agency 4% (4) Intergovernmental or international organisation: 10% (10) Intergovernmental or international organisati… 10% (10) University or research institute: 40% (41) University or research institute 40% (41) NGO, think tank or industry association: 15% (15) NGO, think tank or industry association 15% (15) Consultancy: 7% (7) Consultancy 7% (7)
Regions of expertise n=103
Europe: 49% Europe 49% Global / no specific region: 31% Global / no specific region 31% Australia and New Zealand: 23% Australia and New Zealand 23% China: 21% China 21% India and South-East Asia: 14% India and South-East Asia 14% Japan and South Korea: 11% Japan and South Korea 11% Middle East and North Africa: 10% Middle East and North Africa 10% Brazil and Latin America: 9% Brazil and Latin America 9% North America: 7% North America 7% Sub-Saharan Africa: 0% Sub-Saharan Africa 0%

Share of respondents selecting each option.

Module 1 · 104 experts

Cross-cutting outlook

Hydrogen-based direct reduction is expected to be real but slow: a small share of steel by 2040, held back by offtake and energy prices more than by technology.

  • Steel in 2040: unabated BF-BOF 42%, scrap EAF 30%, natural-gas DRI-EAF 12%, hydrogen DRI-EAF 6% (medians; Q8, n=88).
  • Even odds of 10 Mt a year of hydrogen DRI capacity by 2037; the pooled curve passes 25% in 2033 and 75% in 2043 Q9, n=89.
  • In 2030, 74% expect hydrogen steel to cost at least 25% more than BF-BOF steel; by 2040, 82% expect a premium of 25% or less Q10, n=84.
  • Barriers to investment: offtake and green-premium uncertainty (named by 80%) and hydrogen or electricity price risk (70%); first-of-a-kind technology risk by only 13% Q11, n=98.
  • For the next five years experts favour lead markets or quotas, then capital grants and carbon contracts for difference Q12, n=92; 64% expect new DRI projects to start on natural gas Q16, n=92.
Q8 Allocate global crude steel production in 2040 across production routes (percent, summing to 100). n=88
0%20%40%60%80%100% BF-BOF, unabated: median 42.0%, IQR 30.0–50.0, range 5.0–71.0 (n=88) BF-BOF, unabated Reference: 70% 42 BF-BOF with CCUS and/or hydrogen injection: median 5.0%, IQR 2.8–10.0, range 0.0–32.0 (n=88) BF-BOF with CCUS and/or hydrogen in… Reference: 0% 5 Natural-gas DRI-EAF, with or without CCS: median 12.0%, IQR 10.0–15.0, range 4.0–30.0 (n=87) Natural-gas DRI-EAF, with or withou… Reference: 7% 12 Hydrogen-based DRI-EAF, including imported green iron: median 5.6%, IQR 3.0–10.0, range 0.0–30.0 (n=88) Hydrogen-based DRI-EAF, including i… Reference: 0% 6 Scrap-based EAF: median 30.0%, IQR 25.0–35.0, range 10.0–60.0 (n=88) Scrap-based EAF Reference: 22% 30 Other novel routes (electrolysis, smelting reduction, plasma): median 2.0%, IQR 1.0–3.0, range 0.0–10.0 (n=79) Other novel routes (electrolysis, s… Reference: 1% 2

Box: interquartile range · thick line: median · whiskers: min–max · number: median · purple line: reference value

Q11 Rank the three biggest barriers to reaching final investment decision on commercial-scale hydrogen-based ironmaking projects today. n=98
Offtake and green-premium uncertainty: 63% · 1st 44% Offtake and green-premium uncertainty 63% · 1st 44% Hydrogen or electricity price risk: 53% · 1st 32% Hydrogen or electricity price risk 53% · 1st 32% Durability and predictability of policy support: 20% · 1st 4% Durability and predictability of policy suppo… 20% · 1st 4% Carbon price and CBAM implementation uncertainty: 18% · 1st 7% Carbon price and CBAM implementation uncertai… 18% · 1st 7% Cost of capital and balance-sheet constraints: 13% · 1st 4% Cost of capital and balance-sheet constraints 13% · 1st 4% First-of-a-kind technology risk: 9% · 1st 3% First-of-a-kind technology risk 9% · 1st 3% Capex escalation and EPC capacity: 8% · 1st 3% Capex escalation and EPC capacity 8% · 1st 3% Permitting and grid-connection delays: 7% · 1st 1% Permitting and grid-connection delays 7% · 1st 1% Security of DR-grade ore supply: 7% · 1st 1% Security of DR-grade ore supply 7% · 1st 1% Skilled workforce availability: 2% · 1st 1% Skilled workforce availability 2% · 1st 1%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

Read the full analysis

Asked to allocate global crude steel production in 2040 across routes, the median expert gave 42% to unabated BF-BOF (IQR 30 to 50%, against roughly 70% today), 30% to scrap-based EAF (IQR 25 to 35%), 12% to natural-gas DRI-EAF with or without CCS (IQR 10 to 15%), 6% to hydrogen-based DRI-EAF including imported green iron (IQR 3 to 10%), 5% to BF-BOF with CCUS or hydrogen injection (IQR 3 to 10%) and 2% to other novel routes such as electrolysis, smelting reduction and plasma (IQR 1 to 3%) (Q8, n=88). This was among the most tightly clustered answers in the elicitation, and the medians barely moved as the sample grew. Experts put even odds on at least 10 Mt a year of hydrogen-based DRI capacity being in commercial operation in 2037 (median of the individual 50% years; IQR 2035 to 2040), with 10% years clustering at 2030 to 2031 and 90% years at 2042. The pooled cumulative curve reaches 25% in 2033, 50% in 2037 and 75% in 2043, and one expert answered never (Q9, n=89).

On cost, 46% expect hydrogen DRI-EAF crude steel in 2030 to carry a 25 to 50% production-cost premium over BF-BOF in the best-in-class location, excluding carbon pricing and subsidies, and 27% expect more than 50%. The centre of the distribution moves one band per five years: in 2035, 42% say 10 to 25% and 35% say 25 to 50%; in 2040, 40% say 10 to 25%, 37% say 0 to 10% and 5% expect hydrogen steel to be cheaper (Q10, n=84). The two barriers to final investment decision that stand out are offtake and green-premium uncertainty (63% of the maximum Borda score; ranked first by 44%) and hydrogen or electricity price risk (53%; first by 32%); the gap between them widened slightly with the later responses, which leaned towards offtake. Durability and predictability of policy support (20%) and carbon price and CBAM implementation uncertainty (18%) form a second tier; first-of-a-kind technology risk (9%), permitting and grid delays (7%), security of DR-grade ore supply (7%) and skilled workforce availability (2%) trail (Q11, n=98).

Policy preferences over the next five years are broad rather than concentrated, and the extra responses reshuffled the top of the list without changing its character. Lead markets or quotas for near-zero steel now lead on aggregate score (33% of maximum Borda score; first by 16%), just ahead of capital grants for first plants (28%), which, together with domestic carbon pricing, are the instruments most often ranked first (17% each). Carbon contracts for difference (25%; first by 14%) and domestic carbon pricing (23%; first by 17%) follow, then hydrogen production tax credits or subsidies (21%), green public procurement (17%), border carbon adjustment (16%) and bilateral trade agreements (16%). Harmonised definitions and certification scored lowest (4%) as an accelerator, although the markets module shows experts value it highly as a safeguard (Q12, n=92). Asked what share of hydrogen-based DRI/HBI output will be traded internationally as merchant iron in 2040 rather than consumed on site, the median best estimate was 30% (IQR 20 to 50%; individual estimates from 2% to 92%), one of the widest relative spreads in the whole elicitation (Q13, n=71).

On definitions, 52% want Task 52 to define near-zero emission iron and steel by a technology-neutral emissions-intensity threshold, 31% by a threshold combined with route eligibility rules, 13% prefer no single definition and multiple reported metrics, and 4% would define it by eligible routes (Q14, n=96). The boundary should be Scope 1 and 2 plus upstream Scope 3 (mining, ore processing and transport) for 62%, Scope 1 and 2 only for 26% and the full life cycle including downstream use for 11% (Q15, n=94). Finally, 64% expect the commercial-scale DRI projects reaching final investment decision in 2026 to 2028 to start on natural gas with a planned transition to hydrogen, 33% expect a mix differing by region and only two experts expect hydrogen from commissioning (Q16, n=92).

All 9 questions in this module

Module 2 · 47 experts

Technology readiness and process integration

Hydrogen shaft furnaces are close to commercial, but the first full-scale plant has slipped and hydrogen supply is the binding constraint.

  • Shaft-furnace DRI on 100% hydrogen is rated TRL 7 to 8 by 71%; natural-gas DRI with hydrogen blending is TRL 9 for 59% Q17, n=41.
  • First hydrogen shaft furnace of 1 Mt a year: 49% expect one by 2028 or already, 41% say 2029 to 2032 Q18, n=39.
  • A best-in-class plant in 2035 needs 4.0 MWh of electricity per tonne of crude steel (IQR 3.5 to 4.5; Q20, n=33).
  • Hydrogen supply variability and buffering is the top technical barrier to scale, ranked first by 63% Q21, n=41.
  • A first-of-a-kind plant is expected to run at 65% of nameplate over its first two years (IQR 55 to 70%; Q25, n=33), and 65% expect natural gas to remain the main carbon source in hydrogen DRI in 2035 Q26, n=43.
Q17 What is the current technology readiness of each route? n=41
Shaft-furnace DRI on 100% hydrogen, DR-… Shaft-furnace DRI on 100% hydrogen, DR-grade pellets · TRL 5 to 6: 20% (n=41)Shaft-furnace DRI on 100% hydrogen, DR-grade pellets · TRL 7 to 8: 71% (n=41)Shaft-furnace DRI on 100% hydrogen, DR-grade pellets · TRL 9 or above: 10% (n=41) Mean score (n=41)7.3 Shaft-furnace DRI on natural gas with 3… Shaft-furnace DRI on natural gas with 30% or more hydrogen blending · TRL 5 to 6: 2% (n=41)Shaft-furnace DRI on natural gas with 30% or more hydrogen blending · TRL 7 to 8: 39% (n=41)Shaft-furnace DRI on natural gas with 30% or more hydrogen blending · TRL 9 or above: 59% (n=41) Mean score (n=41)8.6 Fluidised-bed hydrogen reduction of fin… Fluidised-bed hydrogen reduction of fines · TRL 1 to 4: 13% (n=39)Fluidised-bed hydrogen reduction of fines · TRL 5 to 6: 51% (n=39)Fluidised-bed hydrogen reduction of fines · TRL 7 to 8: 36% (n=39) Mean score (n=39)5.9 Hydrogen DRI plus electric smelting fur… Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 1 to 4: 10% (n=40)Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 5 to 6: 48% (n=40)Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 7 to 8: 38% (n=40)Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 9 or above: 5% (n=40) Mean score (n=40)6.2 Hydrogen plasma or smelting reduction Hydrogen plasma or smelting reduction · TRL 1 to 4: 45% (n=38)Hydrogen plasma or smelting reduction · TRL 5 to 6: 55% (n=38) Mean score (n=38)4.4 Molten oxide electrolysis Molten oxide electrolysis · TRL 1 to 4: 47% (n=38)Molten oxide electrolysis · TRL 5 to 6: 42% (n=38)Molten oxide electrolysis · TRL 7 to 8: 11% (n=38) Mean score (n=38)4.5 Low-temperature aqueous iron electrowin… Low-temperature aqueous iron electrowinning · TRL 1 to 4: 47% (n=38)Low-temperature aqueous iron electrowinning · TRL 5 to 6: 45% (n=38)Low-temperature aqueous iron electrowinning · TRL 7 to 8: 8% (n=38) Mean score (n=38)4.5 Blast furnace with hydrogen injection Blast furnace with hydrogen injection · TRL 1 to 4: 5% (n=41)Blast furnace with hydrogen injection · TRL 5 to 6: 20% (n=41)Blast furnace with hydrogen injection · TRL 7 to 8: 44% (n=41)Blast furnace with hydrogen injection · TRL 9 or above: 32% (n=41) Mean score (n=41)7.5 BF-BOF with CCUS BF-BOF with CCUS · TRL 1 to 4: 7% (n=41)BF-BOF with CCUS · TRL 5 to 6: 29% (n=41)BF-BOF with CCUS · TRL 7 to 8: 54% (n=41)BF-BOF with CCUS · TRL 9 or above: 10% (n=41) Mean score (n=41)6.8 Biomass or biochar-based ironmaking Biomass or biochar-based ironmaking · TRL 1 to 4: 7% (n=41)Biomass or biochar-based ironmaking · TRL 5 to 6: 15% (n=41)Biomass or biochar-based ironmaking · TRL 7 to 8: 39% (n=41)Biomass or biochar-based ironmaking · TRL 9 or above: 39% (n=41) Mean score (n=41)7.7
TRL 1 to 4TRL 5 to 6TRL 7 to 8TRL 9 or aboveNumber at right: mean score
Q18 When will the first commercial-scale plant (at least 1 Mt/yr) be operating for each route? n=39
Shaft-furnace DRI on 100% hydrogen Shaft-furnace DRI on 100% hydrogen · Already operating: 8% (n=39)Shaft-furnace DRI on 100% hydrogen · By 2028: 41% (n=39)Shaft-furnace DRI on 100% hydrogen · 2029 to 2032: 41% (n=39)Shaft-furnace DRI on 100% hydrogen · 2033 to 2036: 3% (n=39)Shaft-furnace DRI on 100% hydrogen · 2037 to 2040: 5% (n=39)Shaft-furnace DRI on 100% hydrogen · After 2040: 3% (n=39) Mean score (n=39)2029.5 Fluidised-bed hydrogen reduction of fin… Fluidised-bed hydrogen reduction of fines · Already operating: 6% (n=36)Fluidised-bed hydrogen reduction of fines · 2029 to 2032: 22% (n=36)Fluidised-bed hydrogen reduction of fines · 2033 to 2036: 56% (n=36)Fluidised-bed hydrogen reduction of fines · 2037 to 2040: 8% (n=36)Fluidised-bed hydrogen reduction of fines · After 2040: 8% (n=36) Mean score (n=36)2034.3 Hydrogen DRI plus electric smelting fur… Hydrogen DRI plus electric smelting furnace · Already operating: 3% (n=37)Hydrogen DRI plus electric smelting furnace · By 2028: 8% (n=37)Hydrogen DRI plus electric smelting furnace · 2029 to 2032: 43% (n=37)Hydrogen DRI plus electric smelting furnace · 2033 to 2036: 27% (n=37)Hydrogen DRI plus electric smelting furnace · 2037 to 2040: 16% (n=37)Hydrogen DRI plus electric smelting furnace · After 2040: 3% (n=37) Mean score (n=37)2032.8 Hydrogen plasma or smelting reduction Hydrogen plasma or smelting reduction · 2029 to 2032: 3% (n=35)Hydrogen plasma or smelting reduction · 2033 to 2036: 20% (n=35)Hydrogen plasma or smelting reduction · 2037 to 2040: 34% (n=35)Hydrogen plasma or smelting reduction · After 2040: 40% (n=35)Hydrogen plasma or smelting reduction · Never: 3% (n=35) Mean score (n=35)2040.1 Molten oxide electrolysis Molten oxide electrolysis · Already operating: 3% (n=35)Molten oxide electrolysis · By 2028: 3% (n=35)Molten oxide electrolysis · 2033 to 2036: 14% (n=35)Molten oxide electrolysis · 2037 to 2040: 20% (n=35)Molten oxide electrolysis · After 2040: 46% (n=35)Molten oxide electrolysis · Never: 14% (n=35) Mean score (n=35)2040.5 Low-temperature iron electrowinning Low-temperature iron electrowinning · 2029 to 2032: 3% (n=35)Low-temperature iron electrowinning · 2033 to 2036: 26% (n=35)Low-temperature iron electrowinning · 2037 to 2040: 17% (n=35)Low-temperature iron electrowinning · After 2040: 49% (n=35)Low-temperature iron electrowinning · Never: 6% (n=35) Mean score (n=35)2040.5
Already operatingBy 20282029 to 20322033 to 20362037 to 2040After 2040NeverNumber at right: mean score
Read the full analysis

Forty-seven experts chose this module, eight more than in the first cut, and the readiness picture barely moved. Shaft-furnace DRI on natural gas with 30% or more hydrogen blending is rated TRL 9 or above by 59% and TRL 7 to 8 by 39%. Shaft-furnace DRI on 100% hydrogen with DR-grade pellets sits at TRL 7 to 8 for 71% (20% say TRL 5 to 6, 10% say TRL 9). Blast furnaces with hydrogen injection (44% TRL 7 to 8, 32% TRL 9) and biomass or biochar-based ironmaking (39% and 39%) are rated similarly mature; BF-BOF with CCUS is mostly TRL 7 to 8 (54%). Hydrogen DRI with an electric smelting furnace for lower-grade ores is split between TRL 5 to 6 (48%) and TRL 7 to 8 (38%); fluidised-bed hydrogen reduction of fines is mostly TRL 5 to 6 (51%). Hydrogen plasma or smelting reduction (55% TRL 5 to 6, 45% TRL 1 to 4), low-temperature aqueous iron electrowinning (47% TRL 1 to 4) and molten oxide electrolysis (47% TRL 1 to 4) are the least mature (Q17, n=41).

Timing follows readiness, but the extra responses have pushed the first commercial hydrogen shaft furnace a little later. For shaft-furnace DRI on 100% hydrogen, 8% say a plant of at least 1 Mt a year is already operating, 41% expect one by 2028 and an equal 41% in 2029 to 2032, so the group is now evenly split on whether the first full-scale plant arrives this decade or early next (Q18, n=39). Hydrogen DRI plus an electric smelting furnace is expected mostly in 2029 to 2032 (43%) and fluidised-bed reduction mostly in 2033 to 2036 (56%). Hydrogen plasma or smelting reduction (40% after 2040), molten oxide electrolysis (46% after 2040, 14% never) and electrowinning (49% after 2040, 6% never) are not expected at commercial scale before 2040 by most experts. Asked which route will process most low-grade hematite ore into hydrogen-based iron by 2035, 41% chose shaft-furnace DRI on lower-grade pellets followed by an electric smelting furnace and BOF, 23% said low-grade ore will largely not be used for hydrogen-based iron by 2035, 15% chose beneficiation to DR-grade pellets and shaft-furnace DRI-EAF and 10% fluidised-bed reduction of fines (Q19, n=39).

The best-in-class electricity requirement of a hydrogen DRI-EAF plant in 2035, including electrolysis but excluding mining and pelletising, is put at 4.0 MWh per tonne of crude steel (median best estimate; IQR 3.5 to 4.5; individual estimates 2.0 to 6.0; median low 3.3 and median high 5.0), unchanged by the four additional estimates (Q20, n=33). The leading technical barrier to scaling from demonstration to multiple commercial plants is hydrogen supply variability and buffering, with 73% of the maximum Borda score and ranked first by 63%; availability and consistency of DR-grade pellets (29%) and heat balance and gas preheating with hydrogen (26%) follow (Q21, n=41). Among EAF-side challenges, gangue and slag volume from lower-grade DRI is rated most serious (mean 4.3 of 5; median 5), then scrap residuals limiting the product mix (3.5); operating skills and workforce transition is rated least serious (2.5) (Q22, n=36). A second rating of challenges based on public evidence puts hydrogen supply reliability, pressure and purity at the plant boundary first (3.9), then pellet sticking, clustering and fines generation (3.5), electric smelting furnace slag chemistry (3.5) and the behaviour of magnetite or lower-grade ore in fluidised or flash reduction (3.5); DRI reoxidation during handling and transport is rated lowest (2.8) (Q24, n=32).

The data that are hardest to obtain are operating costs and availability of demonstration plants (77%) and realised capex of demonstration and first commercial plants (69%), well ahead of measured energy and hydrogen consumption (38%) and reliability, maintenance and failure data (33%) (Q23, n=39). A first-of-a-kind commercial hydrogen shaft-furnace plant is expected to achieve a 65% capacity factor over its first 24 months (median best estimate; IQR 55 to 70%; median high 80%), although the newer respondents are more pessimistic about the downside, pulling the median low estimate from 50% to 40% (Q25, n=33). Natural gas or methane injection, used pragmatically despite the emissions, is expected by 65% to be the dominant source of carbon added to hydrogen DRI in 2035; 16% expect biochar or other solid biogenic carbon, 14% a mix with no single dominant source and 5% biogenic CO2 or biomethane (Q26, n=43). Where an electric smelting furnace feeds flat, high-quality steel, 41% expect the BOF to take most tonnage by 2035 with smelter hot metal treated like blast-furnace hot metal, 36% expect the EAF with hot metal blended with scrap, and 15% expect a regional split (Q27, n=39). At co-located sites commissioned in 2028 to 2035, 71% expect hot DRI charged directly to the melter with no HBI step and 18% hot-transport-vessel charging with partial cooling (Q28, n=34). Water consumption of a best-in-class plant in 2035, including electrolysis feed water and cooling, is put at 8.7 cubic metres per tonne of crude steel (median best estimate; IQR 3.9 to 13; n=16), a small sample whose spread widened with the three new answers and is now among the most dispersed numeric answers in the elicitation (Q29).

All 13 questions in this module

Module 3 · 34 experts

Process flexibility

Hydrogen-ready plants are expected to flex with gas prices and seasons rather than run on hydrogen alone, and flexibility is worth most where power is cheap.

  • Minimum stable rate of a hydrogen shaft furnace in 2030: 50 to 70% of nameplate for 35% of experts, 70 to 90% for another 35% Q30, n=20.
  • In a least-cost 2035 system, flexibility sits mainly with electrolyser modulation and hydrogen storage, 25 points each of 100 Q32, n=20.
  • The most feasible feed flexibility is varying the scrap-to-DRI ratio with market prices, rated 4.5 of 5 Q34, n=21.
  • Running mostly on natural gas with hydrogen only when subsidised scores highest for 2035 (56% of maximum); varying the hydrogen share by season or contract is named by 89% Q35, n=27.
  • Locational flexibility (49% of maximum) and ore-grade flexibility (45%) matter most for competitiveness Q36, n=28.
Q35 How will most hydrogen-ready DRI plants use their reductant flexibility in 2035? Rank the three most likely modes of operation, most likely first. n=27
Run mostly on natural gas with hydrogen used only when subsidised: 56% · 1st 37% Run mostly on natural gas with hydrogen used … 56% · 1st 37% Vary the hydrogen share seasonally or by contract period: 53% · 1st 19% Vary the hydrogen share seasonally or by cont… 53% · 1st 19% Run a fixed hydrogen-natural gas blend set by certification thresholds: 37% · 1st 11% Run a fixed hydrogen-natural gas blend set by… 37% · 1st 11% Run on hydrogen only; natural gas kept for start-up and back-up: 31% · 1st 22% Run on hydrogen only; natural gas kept for st… 31% · 1st 22% Switch between hydrogen and natural gas from hour to hour with prices: 23% · 1st 11% Switch between hydrogen and natural gas from … 23% · 1st 11%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

Q36 Rank the three forms of flexibility that will matter most for the competitiveness of hydrogen-based ironmaking. n=28
Locational flexibility (moving reduction to where power is cheap): 49% · 1st 36% Locational flexibility (moving reduction to w… 49% · 1st 36% Ore grade and feed flexibility: 45% · 1st 21% Ore grade and feed flexibility 45% · 1st 21% Hydrogen and natural-gas reductant switching: 40% · 1st 21% Hydrogen and natural-gas reductant switching 40% · 1st 21% Electrical load flexibility (turndown and ramping): 32% · 1st 11% Electrical load flexibility (turndown and ram… 32% · 1st 11% Scrap-ratio flexibility in the EAF: 15% · 1st 7% Scrap-ratio flexibility in the EAF 15% · 1st 7% Output and inventory flexibility (stockpiling HBI): 10% · 1st 0% Output and inventory flexibility (stockpiling… 10% · 1st 0% Product flexibility (hot DRI, HBI, hot metal via ESF): 8% · 1st 4% Product flexibility (hot DRI, HBI, hot metal … 8% · 1st 4%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

Read the full analysis

Thirty-four experts chose this module, and several questions drew fewer than twenty-five answers, so results here are indicative. The minimum stable production rate a hydrogen shaft furnace will sustain in 2030 without damaging refractories or product quality is put at 50 to 70% of nameplate by 35% and 70 to 90% by another 35%; 15% see no meaningful turndown (Q30, n=20). Moving between 50% and 100% of nameplate is expected to take 6 to 24 hours by 36% and 1 to 6 hours by 32%; 18% say more than 24 hours and 9% under an hour (Q31, n=22).

Asked where flexibility should sit in a least-cost renewable-powered hydrogen DRI system in 2035, the median allocation gave the largest shares to electrolyser modulation (median 25 points; IQR 13.5 to 34.5) and hydrogen storage (25; IQR 10 to 29), which are now exactly level; batteries or other electricity storage (11; IQR 10 to 25), DRI plant load modulation, grid imports or firm power contracts and EAF scheduling (10 each) form a second tier, with DRI/HBI inventory buffering at 7.5 (Q32, n=20). By 2032, experts are evenly split between grid-connected electrolysers running near-continuously and decoupled from on-site renewables (38%) and hydrogen storage between electrolyser and DRI plant (38%) as the more common integration architecture, with 21% expecting battery buffering of electrolyser output; a single expert expects direct coupling with the DRI plant absorbing the variability (Q37, n=24).

On feed, the lowest pellet iron content a hydrogen shaft furnace will use economically by 2030, assuming an electric smelting furnace or adapted EAF downstream, is 62 to 65% Fe for 46% and 65 to 67% Fe for 38% (Q33, n=24). Among feed-flexibility capabilities at commercial scale by 2035, varying the scrap-to-DRI ratio in the EAF with market prices is rated most feasible (mean 4.5 of 5), followed by accepting higher gangue with an electric smelting furnace (3.8), switching pellet suppliers without process changes (3.6), running on magnetite concentrates as well as hematite (3.5) and blending DR-grade and BF-grade pellets (3.5); charging lump ore alongside pellets in a hydrogen shaft furnace is rated least feasible (3.4) (Q34, n=21).

How hydrogen-ready plants will actually use their reductant flexibility in 2035 divided the group, and the extra responses tilted it towards gas. Running mostly on natural gas with hydrogen used only when subsidised now has the highest Borda score (56% of maximum; ranked first by 37%), just ahead of varying the hydrogen share seasonally or by contract period (53%; mentioned by 89%, more than any other). A fixed hydrogen-natural gas blend set by the plant design (37%) has overtaken hydrogen-only operation with natural gas kept for start-up and back-up (31%; first by 22%), which has slipped to fourth (Q35, n=27). Among forms of flexibility that matter for competitiveness, locational flexibility, moving reduction to where power is cheap, still leads on Borda score (49%; first by 36%), with ore grade and feed flexibility close behind (45%; first by 21%, and mentioned by 75%, more than any other) and ahead of hydrogen and natural-gas reductant switching (41%) and electrical load flexibility (32%) (Q36, n=28).

All 8 questions in this module

Module 4 · 37 experts

Resources, supply chains and infrastructure

Green iron will be made where firmed renewable power is cheapest and shipped to the steelmaker; DR-grade feed and the size of the merchant market are the open questions.

  • Hydrogen-based iron production in 2040: China 20.5%, Middle East and North Africa 15%, Europe 12%, Australia 10% (medians; Q38, n=30).
  • Tightest bottlenecks for a first 2 Mt a year export project: the cost of firmed renewable power (named by 74%), finance (62%) and offtake certainty (59%) Q40, n=34.
  • Largest corridors in 2040: Middle East and North Africa to Europe (60%), Australia to China (50%), Australia to Japan and Korea (43%) Q43, n=30.
  • Merchant hydrogen-based iron traded in 2035: 35 Mt a year (IQR 12.5 to 60), one of the widest spreads Q44, n=24; a liquid HBI market is expected around 2040 Q53, n=22.
  • 71% expect beneficiated magnetite pellets to supply most traded green iron in 2035 Q48, n=28; 65% Fe is the feed grade below which a shaft furnace cannot run without further upgrading Q50, n=17.
Q38 Allocate global hydrogen-based DRI/HBI production in 2040 across producing regions (percent, summing to 100). n=30
0%20%40%60%80%100% Australia: median 10.0%, IQR 7.3–20.0, range 0.0–40.0 (n=28) Australia 10 Middle East and North Africa: median 15.0%, IQR 10.0–22.9, range 0.0–44.0 (n=27) Middle East and North Africa 15 Brazil and rest of Latin America: median 10.0%, IQR 5.0–11.1, range 0.0–25.7 (n=25) Brazil and rest of Latin America 10 Sub-Saharan Africa: median 5.0%, IQR 1.0–5.0, range 0.0–11.1 (n=17) Sub-Saharan Africa 5 China: median 20.5%, IQR 14.8–32.0, range 5.0–64.0 (n=28) China 21 India: median 6.0%, IQR 4.0–13.5, range 0.0–21.2 (n=23) India 6 Japan, Korea and South-East Asia: median 5.0%, IQR 2.5–10.2, range 0.0–20.0 (n=23) Japan, Korea and South-East Asia 5 Europe: median 11.6%, IQR 6.5–25.0, range 0.0–75.0 (n=28) Europe 12 North America: median 10.0%, IQR 5.1–10.0, range 0.0–25.0 (n=26) North America 10 Other regions: median 1.0%, IQR 0.0–5.0, range 0.0–11.1 (n=17) Other regions 1

Box: interquartile range · thick line: median · whiskers: min–max · number: median

Q43 Which green iron trade corridors will carry the largest volumes in 2040? Select up to three. n=30
Middle East and North Africa to Europe: 60% Middle East and North Africa to Europe 60% Australia to China: 50% Australia to China 50% Australia to Japan and Korea: 43% Australia to Japan and Korea 43% Mostly domestic within China: 40% Mostly domestic within China 40% Brazil to Europe: 27% Brazil to Europe 27% Middle East to South and East Asia: 23% Middle East to South and East Asia 23% Australia to Europe: 17% Australia to Europe 17% Brazil to North-East Asia: 13% Brazil to North-East Asia 13% Mostly domestic within India: 10% Mostly domestic within India 10% Sub-Saharan Africa to Europe: 0% Sub-Saharan Africa to Europe 0% Canada to Europe: 0% Canada to Europe 0%

Share of respondents selecting each option.

Read the full analysis

Thirty-seven experts chose this module. Allocating global hydrogen-based DRI/HBI production in 2040 across regions, the median expert gave China 20.5% (IQR 15 to 32), the Middle East and North Africa 15% (IQR 10 to 23), Europe 12% (IQR 6.5 to 25), Australia 10% (IQR 7 to 20), Brazil and Latin America 10%, North America 10%, India 6%, Japan, Korea and South-East Asia 5%, Sub-Saharan Africa 5% and other regions 1% (Q38, n=30). China's median edged down with the new responses, and the gap between it and the next region has narrowed. Only 10 to 25% of today's seaborne iron ore trade is seen as economically upgradeable to DR-grade feed by 2035 by 44%; 33% say less than 10% and 22% say 25 to 50%, so more than three in four still put the upgradeable share at a quarter or less (Q39, n=27).

For a first 2 Mt a year green iron export project in a best-in-class location, the ordering of bottlenecks has changed with the seven additional rankings. The cost of firmed renewable electricity is now the tightest bottleneck (50% of maximum Borda; ranked first by 26%; mentioned by 74%), followed by offtake certainty (45%; also first by 26%) and finance and cost of capital (39%; first by 15%; mentioned by 62%), which had led the first cut. Electrolyser cost and delivery (23%), DR-grade ore or beneficiation capacity (19%) and transmission and grid connection (19%) follow; permitting and workforce were not ranked by anyone (Q40, n=34). Public or shared investment is rated most important for high-voltage transmission (mean 4.3 of 5) and hydrogen pipelines and storage (4.3), then water supply and desalination (3.9) and port and HBI handling (3.7); pellet and beneficiation plants (2.8) and workforce housing (2.5) are seen as private matters (Q41, n=26). Roughly three in five agree or strongly agree that most new hydrogen-based iron capacity by 2035 will be built in vertically integrated hubs co-locating ore, renewable electricity, hydrogen and DRI (42% agree, 18% strongly agree, 30% neutral, 9% disagree); the newer respondents were more often neutral than the first group (Q42, n=33).

The corridors expected to carry the largest green iron volumes in 2040 are the Middle East and North Africa to Europe (60%) and Australia to China (50%), then Australia to Japan and Korea (43%) and mostly domestic trade within China (40%), Brazil to Europe (27%) and the Middle East to South and East Asia (23%); Sub-Saharan Africa to Europe and Canada to Europe were not chosen by anyone (Q43, n=30). The volume of merchant hydrogen-based iron traded internationally in 2035 is put at 35 Mt a year (median best estimate; IQR 12.5 to 60; individual estimates from 4 to 300 Mt), still among the most dispersed answers in the elicitation (Q44, n=24). The cost of firmed renewable electricity is the dominant locational driver over the next decade (mean 4.7 of 5; 23 of 32 rated it 5 and nobody below 4), ahead of political and trade risk (4.2), capital and operating subsidies (4.1), carbon pricing and border-adjustment exposure (3.9), access to low-cost natural gas as a bridge (3.8) and proximity to green steel demand (3.6); existing steel plants and skilled workforce matter least (3.0) (Q45, n=32).

On DR-grade feed, the 2030 gap between the pellet demand of operating DRI plants and the supply actually available is put at 25 Mt a year (median best estimate; IQR 16 to 54; n=16, with answers from 0 to 100 Mt and high disagreement) (Q46). If high-grade ore supply grew by an additional 100 Mt a year above 65% Fe by 2030, opinion is now spread fairly evenly: 28% expect modest easing, 24% material easing of DR-grade availability and 24% say the effect depends mainly on beneficiation and pelletising capacity, while 14% expect availability still to tighten and 10% little change; the first cut had leaned towards "depends on beneficiation", and the newer respondents are more optimistic (Q47, n=29). Magnetite beneficiated to DR-grade pellets and fed to shaft-furnace hydrogen DRI is expected by 71% to supply most traded green iron in 2035; 11% expect global green iron trade to stay below 5 Mt a year (Q48, n=28). The probability that more than half of global DRI output in 2035 comes from ore below 65% Fe as mined is put at 40% (median; IQR 22.5 to 57.5%) (Q49, n=27). The feed grade below which a shaft furnace can no longer operate commercially without further upgrading is 65% Fe (median best estimate; IQR 62 to 65; n=17), still the most tightly agreed numeric answer in the elicitation (Q50), and the cost premium for producing DR-grade pellets is USD 35 per tonne of pellets (median; IQR 28.5 to 50; n=18) (Q51).

Among constraints on capacity additions in 2026 to 2032, no single item stands out: electrolyser manufacturing capacity and delivery times (mean 3.6 of 5), high-voltage transformers and grid equipment (3.6) and shaft-furnace and DRI-plant EPC and equipment capacity (3.5) are rated most serious, electric smelting furnace equipment (3.4) close behind; specialist construction labour (3.2) and graphite electrodes (3.0) least (Q52, n=25). A liquid merchant market for hydrogen-based HBI, with multiple independent buyers and sellers trading against a published benchmark, is expected around 2040 (median 50% year 2040; IQR 2037 to 2042; pooled curve crosses 50% in 2040 and 75% in 2045; nobody answered never) (Q53, n=22).

All 16 questions in this module

Module 5 · 26 experts

Projects, finance and offtake

The smallest module, with 26 experts: government revenue support is what gets projects to investment decision, and sponsors carry most of the risk today.

  • Cost of debt for a first-of-a-kind 2 Mt a year project without government guarantees: 9% a year (IQR 8 to 12.6; Q54, n=12).
  • Government offtake or revenue-support contracts are the mechanism most expected to lift projects to investment decision by 2029, ranked first by 62%; buyer-backed offtake usable as collateral is second Q55, n=21.
  • In today’s financings the sponsor bears most hydrogen and electricity price risk (60%) and most technology and completion risk (71%) Q56, n=15.
  • By 2030 a fixed premium over a benchmark index is the most expected offtake structure; 47% still rank bespoke deals among their top three Q58, n=15.
Q55 Rank the three financing or risk-sharing mechanisms that would most increase the number of hydrogen-based iron and steel projects reaching final investment decision by 2029. n=21
Government offtake or revenue-support contracts (carbon contracts for difference): 76% · 1st 62% Government offtake or revenue-support contrac… 76% · 1st 62% Buyer-backed offtake agreements usable as loan collateral: 43% · 1st 19% Buyer-backed offtake agreements usable as loa… 43% · 1st 19% Direct capital grants or public equity co-investment: 33% · 1st 14% Direct capital grants or public equity co-inv… 33% · 1st 14% Sovereign or strategic-investor equity stakes: 17% · 1st 5% Sovereign or strategic-investor equity stakes 17% · 1st 5% Multilateral development bank guarantees or first-loss capital: 8% · 1st 0% Multilateral development bank guarantees or f… 8% · 1st 0% Blended or concessional finance from climate funds: 8% · 1st 0% Blended or concessional finance from climate … 8% · 1st 0% Export credit agency-backed debt: 6% · 1st 0% Export credit agency-backed debt 6% · 1st 0% Insurance covering technology or completion risk: 6% · 1st 0% Insurance covering technology or completion r… 6% · 1st 0% Green or sustainability-linked bond markets: 2% · 1st 0% Green or sustainability-linked bond markets 2% · 1st 0%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

Q56 In a typical hydrogen DRI project financing today, which party bears most of each risk? n=15
Offtake and demand risk Offtake and demand risk · Project sponsor or equity: 53% (n=15)Offtake and demand risk · Government support scheme: 7% (n=15)Offtake and demand risk · Buyer, via offtake terms: 33% (n=15)Offtake and demand risk · Lender or debt: 7% (n=15) Mean score (n=15)n=15 Hydrogen and electricity price risk Hydrogen and electricity price risk · Project sponsor or equity: 60% (n=15)Hydrogen and electricity price risk · Government support scheme: 7% (n=15)Hydrogen and electricity price risk · Buyer, via offtake terms: 7% (n=15)Hydrogen and electricity price risk · Largely unallocated or uninsured: 27% (n=15) Mean score (n=15)n=15 Technology and completion risk Technology and completion risk · Project sponsor or equity: 71% (n=14)Technology and completion risk · Government support scheme: 14% (n=14)Technology and completion risk · Lender or debt: 7% (n=14)Technology and completion risk · Largely unallocated or uninsured: 7% (n=14) Mean score (n=14)n=14 Policy durability risk, such as subsidy… Policy durability risk, such as subsidy or contract survival · Project sponsor or equity: 47% (n=15)Policy durability risk, such as subsidy or contract survival · Government support scheme: 27% (n=15)Policy durability risk, such as subsidy or contract survival · Buyer, via offtake terms: 7% (n=15)Policy durability risk, such as subsidy or contract survival · Largely unallocated or uninsured: 20% (n=15) Mean score (n=15)n=15 Political and cross-border trade risk Political and cross-border trade risk · Project sponsor or equity: 40% (n=15)Political and cross-border trade risk · Government support scheme: 27% (n=15)Political and cross-border trade risk · Buyer, via offtake terms: 13% (n=15)Political and cross-border trade risk · Largely unallocated or uninsured: 20% (n=15) Mean score (n=15)n=15
Project sponsor or equityGovernment support schemeBuyer, via offtake termsLender or debtLargely unallocated or uninsuredNumber at right: mean score
Read the full analysis

Twenty-six experts chose this module, the smallest group, and most questions drew 12 to 21 answers; treat these as indicative. The cost of debt a first-of-a-kind 2 Mt a year hydrogen DRI project with no operating track record would face at final investment decision in 2026 or 2027, without government guarantees, is put at 9% a year (median best estimate; IQR 8 to 12.6; individual estimates 3 to 20%; n=12) (Q54). Government offtake or revenue-support contracts such as carbon contracts for difference are by far the mechanism most expected to increase the number of projects reaching final investment decision by 2029 (76% of maximum Borda; ranked first by 62%; mentioned by 86%). Buyer-backed offtake agreements usable as loan collateral (43%; mentioned by 62%) have moved into second place ahead of direct capital grants or public equity co-investment (33%). Sovereign or strategic-investor equity stakes (17%), export credit, concessional climate finance, development-bank guarantees, insurance and green bonds each scored under 20% (Q55, n=21).

In a typical hydrogen DRI project financing today, the project sponsor or equity is seen as bearing most of the hydrogen and electricity price risk (60%; 27% say it is largely unallocated or uninsurable) and most technology and completion risk (71%; 14% government support scheme). Offtake and demand risk is split between the sponsor (53%) and the buyer via offtake terms (33%). Policy durability risk is spread across the sponsor (47%), government (27%) and unallocated (20%), as is political and cross-border trade risk (40%, 27%, 20%, with 13% on the buyer) (Q56, n=15). Typical 2026 premiums for near-zero or low-emission steel, judged from public assessments, are highest for automotive OEM long-term contracts (46% say USD 120 to 200 per tonne and 31% USD 50 to 120) and lowest for construction and infrastructure (46% no premium, 38% up to USD 50) and the spot market (55% no premium) (Q57, n=13).

By 2030 the most common structure for large green iron and steel offtake agreements is expected to be a fixed premium over a benchmark index for the contract life (53% of maximum Borda; first by 33%), then a government contract for difference covering the gap with the buyer paying the reference price (40%); almost half of experts rank no standard structure, with deals staying bespoke, among their top three (47%), and 27% rank it first, while cost-plus pass-through of the seller's input costs (29%) gained ground with the new responses (Q58, n=15). Of new hydrogen-based DRI capacity reaching operation between 2026 and 2035, experts are now split three ways: 32% expect less than 20% to be brownfield conversions of existing integrated works, 32% expect 20 to 40%, 32% expect 40 to 60% and 5% expect conversion to dominate (Q59, n=19).

All 6 questions in this module

Module 6 · 42 experts

Pathways, competitiveness and power systems

Costs converge on the blast furnace only in the late 2030s, the electricity price is the dominant uncertainty, and scrap-based EAF is the one safe investment.

  • Carbon price for parity with unabated BF-BOF in Europe in 2030, without subsidies: USD 165 per tonne of CO2 (IQR 141 to 200; Q61, n=30).
  • Low-emission hydrogen delivered in 2035: USD 3.0 per kilogram in a best-in-class location (IQR 2.5 to 4.4; Q62, n=34).
  • The renewable electricity price and its firming is the largest source of cost uncertainty, ranked first by 58% Q67, n=36.
  • Even odds in 2040 that European hydrogen steel undercuts unabated BF-BOF with carbon pricing and CBAM but no subsidies Q68, n=27.
  • Scrap-based EAF capacity is the least-regret investment, ranked first by 61% Q72, n=36; a rising, predictable carbon price is the least-regret policy, first by 56% Q75, n=36.
Q61 What carbon price would make hydrogen-based DRI-EAF steel cost-competitive with unabated BF-BOF in Europe in 2030, without other subsidies? n=30

Median low

110.0

Median best

165.0

IQR 141.3–200.0

Median high

232.5

080160240320400Interquartile range of best estimates: 141.3–200.0Median best estimate 165.0Expert: low 75, best 90, high 205Expert: low 75, best 100, high 150Expert: low 80, best 100, high 150Expert: low 60, best 120, high 250Expert: low 75, best 120, high 165Expert: low 90, best 140, high 220Expert: low 100, best 140, high 180Expert: low 110, best 140, high 200Expert: low 90, best 145, high 240Expert: low 90, best 150, high 200Expert: low 100, best 150, high 200Expert: low 100, best 150, high 200Expert: low 100, best 150, high 200Expert: low 110, best 160, high 180Expert: low 150, best 160, high 200Expert: low 100, best 170, high 200Expert: low 120, best 180, high 250Expert: low 165, best 190, high 225Expert: low 120, best 200, high 300Expert: low 120, best 200, high 275Expert: low 150, best 200, high 250Expert: low 150, best 200, high 350Expert: low 150, best 200, high 300Expert: low 180, best 200, high 250Expert: low 120, best 205, high 320Expert: low 100, best 220, high 295Expert: low 180, best 230, high 250Expert: low 130, best 250, high 400Expert: low 150, best 250, high 300Expert: low 200, best 250, high 300

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: USD per tonne CO2

Q72 Which investments are least-regret for a steelmaker deciding today, meaning they hold value across most hydrogen-price and policy scenarios out to 2040? Rank your top three. n=36
Scrap-based EAF capacity and scrap sourcing: 75% · 1st 61% Scrap-based EAF capacity and scrap sourcing 75% · 1st 61% Grid connection and long-term renewable power contracts: 26% · 1st 6% Grid connection and long-term renewable power… 26% · 1st 6% Hydrogen-ready DRI started on natural gas: 25% · 1st 8% Hydrogen-ready DRI started on natural gas 25% · 1st 8% Long-term green-steel offtake contracts with buyers: 23% · 1st 6% Long-term green-steel offtake contracts with … 23% · 1st 6% Electric smelter (ESF/OSBF) capability for lower-grade ore: 13% · 1st 6% Electric smelter (ESF/OSBF) capability for lo… 13% · 1st 6% Keeping blast furnaces running with CCUS or hydrogen-injection readiness: 13% · 1st 6% Keeping blast furnaces running with CCUS or h… 13% · 1st 6% Securing DR-grade ore or beneficiation capacity: 8% · 1st 3% Securing DR-grade ore or beneficiation capaci… 8% · 1st 3% Import agreements for green HBI or merchant iron: 8% · 1st 3% Import agreements for green HBI or merchant i… 8% · 1st 3% Deferring major capital decisions until after 2030: 8% · 1st 3% Deferring major capital decisions until after… 8% · 1st 3%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

Read the full analysis

Forty-two experts chose this module, up from 31 in the previous cut, and the extra answers have sharpened rather than shifted its findings. The domestic cost premium of hydrogen DRI-EAF crude steel over BF-BOF, excluding carbon pricing and subsidies, is put at 25 to 50% in Europe in 2030 by 47% and above 50% by 38%; in Europe in 2035 the centre is still 25 to 50% (44%), with 31% now expecting 10 to 25% and 16% still above 50%. For China the centre is 25 to 50% in 2030 (52%) falling to 10 to 25% in 2035 (45%, with a further 24% at 10% or less) (Q60, n=32). The carbon price that would make hydrogen DRI-EAF steel competitive with unabated BF-BOF in Europe in 2030 without other subsidies is USD 165 per tonne of CO2 (median best estimate; IQR 141 to 200; individual estimates 90 to 250), up from USD 155 in the 84-response cut (Q61, n=30). Low-emission hydrogen delivered to a DRI plant in a best-in-class location in 2035 is expected to cost USD 3.0 per kilogram (median best estimate; IQR 2.5 to 4.4; median low 2.0, median high 4.75). The upper quartile has come down with the extra respondents, but four experts still put their best estimate at USD 8.5 or more, so the spread remains wide (Q62, n=34).

Probabilities for 2030 milestones: China having three or more hydrogen shaft furnaces of 1 Mt a year or larger operating, 70% (median; IQR 50 to 80; n=29); EU CBAM applying to iron and steel at full strength without major derogations, 45% (IQR 25 to 70; n=30); a near-zero steel definition mutually recognised by the EU, Japan and Korea, 45% (n=28); at least 10 Mt a year of hydrogen-based DRI capacity operating globally, 25% (IQR 11.3 to 50; n=34); at least 5 Mt a year of green iron traded between continents, 25% (n=33); a commercial-scale hydrogen DRI plant operating in North America, 20% (IQR 10 to 31.3; n=28) (Q63). Europe is expected to import 10 to 25% of the primary iron units for its near-zero steel as HBI or DRI in 2040 by 39% and 25 to 50% by another 39%; only 6% expect less than 10% and 15% expect more than half (Q64, n=33).

Among power-system issues at multi-million-tonne scale, the cost of firming variable renewables (mean 4.3 of 5), electricity market design and industrial tariffs (4.1) and grid connection lead times (3.9) lead; the value of DRI plants as flexible grid demand (3.0) and DRI plant turndown and ramping (3.0) are rated least important (Q65, n=32). For sustaining hydrogen-based steelmaking after 2030, domestic carbon pricing (4.1), border carbon adjustment (4.1) and carbon contracts for difference (4.0) are rated most effective; plurilateral clubs and sectoral agreements score lowest (3.2) (Q66, n=33). The parameter contributing most to uncertainty in the 2035 cost of hydrogen-based steel is the renewable electricity price and firming (71% of maximum Borda; ranked first by 58%; mentioned by 81%), far ahead of the green premium buyers will pay (29%), carbon price and CBAM (26%) and electrolyser capex and efficiency (21%); the green premium has moved up to second place with the extra respondents, but the three are close and all far behind electricity (Q67, n=36). Hydrogen DRI-EAF steel made in Europe becoming cheaper than unabated BF-BOF steel in Europe, including carbon pricing and border adjustment but excluding subsidies, is given even odds in 2040 (median 50% year; IQR 2036.5 to 2043); the pooled curve crosses 25% in 2035, 50% in 2040 and 75% in 2046 (Q68, n=27).

Global crude steel demand in 2050 is put at 2,025 Mt a year (median best estimate; IQR 1,905 to 2,200; individual estimates 1,200 to 2,600), still the second most tightly agreed numeric answer in the elicitation (Q69, n=30). The delivered price of firmed, near-continuous renewable electricity for a large flexible industrial user in 2030 is expected to be below USD 30 per MWh in China by 43%, with another 38% in the 30 to 50 band; the 30 to 50 band is also the centre for the Middle East and North Africa (45%) and the United States (47%), while Europe is split between 30 to 50 (38%) and 50 or above (52%, with 24% at 70 to 100) (Q70, n=24). Iron and steel is expected to account for 11% of global hydrogen consumption in 2035 (median best estimate; IQR 10 to 18.5) (Q71, n=24).

Three least-regret questions close the module. For a steelmaker deciding today, scrap-based EAF capacity and scrap sourcing is the clear least-regret investment (75% of maximum Borda; first by 61%; mentioned by 89%). Behind it, grid connection and long-term renewable power contracts (26%; mentioned by 44%) and hydrogen-ready DRI started on natural gas (25%) are now almost level, ahead of long-term green-steel offtake contracts (23%); deferring major capital decisions until after 2030 was ranked by 17% and first by one expert (Q72, n=36). For an integrated BF-BOF site in a high-cost jurisdiction facing a reline before 2032, the most common first answer is that it depends entirely on the site (first by 58%), but the most widely mentioned option is to replace the furnace with an EAF fed by scrap and imported HBI (mentioned by 82%; Borda 52%; first by 15%), ahead of hydrogen-ready natural-gas DRI and EAF (24%) and retiring the capacity to source slab or iron elsewhere (23%); relining unabated and deciding later was ranked by 33% but first by only 6% (Q73, n=33). If renewable hydrogen still costs more than USD 4 per kilogram in 2035, scrap-based EAF expansion is judged robust by 85% and at risk by nobody; natural-gas DRI with CCS is robust for 35% and tolerable for another 38%; importing green HBI is tolerable for 63%; hydrogen DRI at the existing steel site is tolerable for 46% and at risk for 46%; the electric smelter on lower-grade ore with hydrogen DRI is at risk for 36%; and the blast furnace with CCUS is at risk of stranding for 58% (Q74, n=28). Among least-regret policies, a rising, predictable carbon price dominates (65% of maximum Borda; first by 56%; mentioned by 72%), then lead markets (29%), grid build-out and faster connection for industrial loads (26%) and border carbon adjustment (24%); research and demonstration funding was ranked by one expert (Q75, n=36).

All 16 questions in this module

Module 7 · 42 experts

Regional pathways

China stays on the blast furnace to 2035, the United States turns to gas DRI, and Japan, Korea and Europe increasingly import their iron units.

  • China in 2035: 62% unabated BF-BOF, 22% scrap EAF, 9% hydrogen DRI (medians; Q76, n=28); whether China becomes a net exporter or importer of hydrogen-based iron is evenly split Q77, n=31.
  • India in 2035: 52% expect coal-based rotary-kiln DRI to remain the dominant direct-reduction route Q78, n=27.
  • United States 2025 to 2035: half expect natural-gas DRI-EAF without CCS to lead new primary capacity; nobody expects hydrogen DRI Q79, n=24.
  • About half of the EU blast-furnace capacity facing a reinvestment decision by 2030 is expected to close permanently by 2035 (median 49%; Q80, n=25).
  • Imported share of primary iron units for near-zero steel by 2050: Japan 65%, South Korea 60%, European Union 50% (medians; Q82, n=32).
Q76 Allocate China's domestic crude steel production in 2035 across production routes (percent, summing to 100). n=28
0%20%40%60%80%100% BF-BOF, unabated: median 62.0%, IQR 50.0–70.0, range 17.0–81.0 (n=28) BF-BOF, unabated 62 BF-BOF with CCUS: median 5.0%, IQR 3.0–8.8, range 1.0–15.0 (n=22) BF-BOF with CCUS 5 Natural-gas DRI-EAF: median 1.0%, IQR 0.0–5.0, range 0.0–27.0 (n=18) Natural-gas DRI-EAF 1 Hydrogen-based DRI-EAF: median 9.0%, IQR 3.0–10.0, range 1.0–70.0 (n=26) Hydrogen-based DRI-EAF 9 Scrap-based EAF: median 22.0%, IQR 15.8–28.5, range 8.0–50.0 (n=28) Scrap-based EAF 22 Other novel routes: median 1.0%, IQR 0.0–2.8, range 0.0–10.0 (n=16) Other novel routes 1

Box: interquartile range · thick line: median · whiskers: min–max · number: median

Q82 And by 2050, what share of each region's primary iron units for near-zero steel will be imported as HBI or DRI rather than reduced domestically? Use the slider for each region. n=32
0%20%40%60%80%100%Japan: median 65%, IQR 50–80 (n=25) Japan 65%South Korea: median 60%, IQR 33–70 (n=23) South Korea 60%European Union: median 50%, IQR 45–60 (n=26) European Union 50%China: median 10%, IQR 5–20 (n=26) China 10%India: median 10%, IQR 5–20 (n=25) India 10%

Dot: median · bar: interquartile range

Read the full analysis

Forty-two experts chose this module, with 24 to 33 answering each question. For China's domestic crude steel in 2035, the median allocation is 62% unabated BF-BOF (IQR 50 to 70), 22% scrap-based EAF (IQR 15.8 to 28.5), 9% hydrogen-based DRI-EAF (IQR 3 to 10), 5% BF-BOF with CCUS and about 1% each for natural-gas DRI-EAF and novel routes; the hydrogen share has slipped from 10% in the previous cut as the newer respondents are more cautious (Q76, n=28). Whether China will be a net exporter or importer of hydrogen-based DRI and HBI in 2035 is finely balanced: 35% expect a modest net exporter, 26% a modest net importer, 16% negligible trade either way and 16% roughly balanced; one expert expects a significant net exporter above 5 Mt a year and one a significant net importer (Q77, n=31).

In India, 52% expect coal-based rotary-kiln DRI to remain the dominant direct-reduction pathway in 2035 and 30% expect BF-BOF expansion to outpace all DRI routes; 11% expect hydrogen-based DRI to dominate new capacity and 7% natural-gas DRI (Q78, n=27). In the United States, a clearer picture has emerged with the larger sample: half of the experts now expect natural-gas DRI-EAF without CCS to account for most new primary ironmaking capacity between 2025 and 2035 (50%), against 29% who expect scrap-EAF to absorb nearly all growth with little new primary capacity and 21% who expect gas DRI with CCS; nobody expects hydrogen DRI or new unabated BF-BOF (Q79, n=24). Of the EU blast-furnace capacity that needs a reinvestment decision by 2030, about half (median best estimate 49%; IQR 25 to 55) is expected to be permanently closed by 2035 with no low-emission replacement on the same site, with the lower quartile falling as more experts put the figure in the 20s (Q80, n=25).

Import dependence for near-zero steel rises over time in the importing regions. By 2035 the median expected share of primary iron units imported as HBI or DRI rather than reduced domestically is 45% for Japan (IQR 10 to 62.5), 35% for South Korea (IQR 5 to 65), 25% for the European Union (IQR 20 to 45), 5% for China and 5% for India (Q81, n=33). By 2050 it is 65% for Japan (IQR 50 to 80), 60% for South Korea (IQR 32.5 to 70), 50% for the European Union (IQR 45 to 60), 10% for China and 10% for India (Q82, n=32). The 2050 answers cluster more tightly than the 2035 ones, and the gap between the two horizons has widened slightly for Europe and Japan with the extra respondents.

All 7 questions in this module

Module 8 · 81 experts

Markets, standards and policy

Buyers will pay a modest premium, automotive moves first, and a shared carbon-accounting rulebook is the gap that most limits decisions.

  • Premium buyers will sustainably pay for near-zero steel in 2030: USD 50 to 100 per tonne for 35%, up to USD 50 for 32%, none for 9% Q83, n=66.
  • Automotive is the only sector most experts expect to move fast by 2035 (56%); construction lags, with 60% expecting slow progress Q84, n=68.
  • Carbon accounting and certification rules are the top knowledge gap, 3.9 of 5 Q85, n=63; every accounting safeguard is rated above 4 of 5, mutual recognition across borders at 4.3 Q86, n=67.
  • One mutually recognised steel carbon-intensity standard: even odds around 2038 on the pooled curve Q87, n=62.
  • The policy most at risk of being weakened by 2030 is the EU ETS free-allocation phase-out timetable, ranked first by 48% Q88, n=52.
Q83 What price premium per tonne will buyers sustainably pay for near-zero emission steel in 2030, averaged across segments? n=66
No premium: 9% (6) No premium 9% (6) Up to USD 50: 32% (21) Up to USD 50 32% (21) USD 50 to 100: 35% (23) USD 50 to 100 35% (23) USD 100 to 200: 18% (12) USD 100 to 200 18% (12) More than USD 200: 6% (4) More than USD 200 6% (4)

Mean score 76.5, median 75.0

Q85 How much does each knowledge gap limit decisions by industry and governments today? n=63
Carbon accounting and certification rul… Carbon accounting and certification rules · 2: 7% (n=60)Carbon accounting and certification rules · 3: 20% (n=60)Carbon accounting and certification rules · 4: 48% (n=60)Carbon accounting and certification rules · 5: 25% (n=60) Mean score (n=60)3.9 Comparison of national policy packages Comparison of national policy packages · 1: 2% (n=57)Comparison of national policy packages · 2: 11% (n=57)Comparison of national policy packages · 3: 33% (n=57)Comparison of national policy packages · 4: 37% (n=57)Comparison of national policy packages · 5: 18% (n=57) Mean score (n=57)3.6 Power-system integration assessment Power-system integration assessment · 1: 5% (n=59)Power-system integration assessment · 2: 12% (n=59)Power-system integration assessment · 3: 29% (n=59)Power-system integration assessment · 4: 32% (n=59)Power-system integration assessment · 5: 22% (n=59) Mean score (n=59)3.5 Plant flexibility and hydrogen bufferin… Plant flexibility and hydrogen buffering requirements · 1: 3% (n=61)Plant flexibility and hydrogen buffering requirements · 2: 10% (n=61)Plant flexibility and hydrogen buffering requirements · 3: 33% (n=61)Plant flexibility and hydrogen buffering requirements · 4: 39% (n=61)Plant flexibility and hydrogen buffering requirements · 5: 15% (n=61) Mean score (n=61)3.5 Operating data from demonstration plants Operating data from demonstration plants · 1: 3% (n=63)Operating data from demonstration plants · 2: 21% (n=63)Operating data from demonstration plants · 3: 19% (n=63)Operating data from demonstration plants · 4: 38% (n=63)Operating data from demonstration plants · 5: 19% (n=63) Mean score (n=63)3.5 Trade and relocation scenarios Trade and relocation scenarios · 1: 2% (n=58)Trade and relocation scenarios · 2: 17% (n=58)Trade and relocation scenarios · 3: 33% (n=58)Trade and relocation scenarios · 4: 29% (n=58)Trade and relocation scenarios · 5: 19% (n=58) Mean score (n=58)3.5 Comparative supply-chain cost model Comparative supply-chain cost model · 2: 21% (n=56)Comparative supply-chain cost model · 3: 25% (n=56)Comparative supply-chain cost model · 4: 39% (n=56)Comparative supply-chain cost model · 5: 14% (n=56) Mean score (n=56)3.5 Processing routes for low-grade ores Processing routes for low-grade ores · 1: 5% (n=62)Processing routes for low-grade ores · 2: 15% (n=62)Processing routes for low-grade ores · 3: 31% (n=62)Processing routes for low-grade ores · 4: 31% (n=62)Processing routes for low-grade ores · 5: 19% (n=62) Mean score (n=62)3.5 Harmonised techno-economic dataset acro… Harmonised techno-economic dataset across routes · 1: 5% (n=61)Harmonised techno-economic dataset across routes · 2: 25% (n=61)Harmonised techno-economic dataset across routes · 3: 34% (n=61)Harmonised techno-economic dataset across routes · 4: 21% (n=61)Harmonised techno-economic dataset across routes · 5: 15% (n=61) Mean score (n=61)3.2 Workforce and skills requirements Workforce and skills requirements · 1: 5% (n=58)Workforce and skills requirements · 2: 29% (n=58)Workforce and skills requirements · 3: 45% (n=58)Workforce and skills requirements · 4: 17% (n=58)Workforce and skills requirements · 5: 3% (n=58) Mean score (n=58)2.8
12345Number at right: mean score
Read the full analysis

Eighty-one experts chose this module, the largest expertise group, and its results are the most stable in the report: no ranking of substance has changed. The premium buyers will sustainably pay for near-zero emission steel in 2030, averaged across segments, is USD 50 to 100 per tonne for 35%, up to USD 50 for 32%, USD 100 to 200 for 18%, no premium for 9% and more than USD 200 for 6% (Q83, n=66). By 2035, automotive is expected to have made rapid progress (40%) or largely transitioned (16%) by a majority; wind, solar and grid equipment follows (36% rapid, 9% largely). Construction and infrastructure (60% slow progress, 22% pilots only), shipbuilding (37% slow, 29% pilots), machinery (44% slow, 27% pilots) and appliances (44% slow) are expected to lag (Q84, n=68).

The knowledge gaps that most limit decisions by industry and governments today are carbon accounting and certification rules (mean 3.9 of 5), comparison of national policy packages (3.6), power-system integration assessment (3.5), plant flexibility and hydrogen buffering requirements (3.5), operating data from demonstration plants (3.5), trade and relocation scenarios (3.5), a comparative supply-chain cost model (3.5) and processing routes for low-grade ores (3.5); workforce and skills requirements limit least (2.8) (Q85, n=63). Every safeguard for emissions-accounting schemes is rated important: transparent eligibility and boundary rules (4.4), prevention of double counting (4.4), mutual recognition between schemes across borders (4.3), public disclosure of the allocation method (4.3), physical traceability or chain of custody (4.2) and independent third-party verification (4.2) (Q86, n=67). A single steel carbon-intensity standard mutually recognised across the EU, Japan, Korea, the United States, India and China is given even odds around 2040 (median 50% year 2040; IQR 2035 to 2047.5; the pooled curve crosses 50% in 2038); four experts answered never for the 50% year and twelve for the 90% year, so the tail of doubt has lengthened with the extra respondents (Q87, n=62).

The policy instrument most likely to be materially weakened, delayed or reversed by 2030 is the EU ETS free-allocation phase-out timetable for 2026 to 2034 (60% of maximum Borda; ranked first by 48%), followed by the EU CBAM extension to downstream products (35%) and the US Section 232 steel tariffs at 50% (35%), which swapped places in the previous cut and are now level to the nearest percentage point; India's green steel procurement mandate (16%) and Germany's carbon contracts for difference (14%) follow (Q88, n=52). On CBAM treating imported DRI and HBI as reportable precursors on verified actual emissions, 47% expect a roughly neutral effect on imported hydrogen-based iron versus EU-produced hydrogen DRI by 2030, 19% say it is too uncertain, 19% expect it to materially favour EU production because the compliance burden is a real barrier and 15% expect it to favour verified low-carbon imports (Q89, n=53).

Carbon contracts for difference are most effective when they cover both the carbon-price gap and the hydrogen or electricity cost gap (76%), run for ten years or more (61%), are combined with demand-side measures such as lead markets or quotas (52%) and sit on a predictable carbon price and free-allocation phase-out path (44%); fast award with limited conditionality was chosen by one expert (Q90, n=54). The EU's hourly renewable-matching requirement for RFNBO hydrogen, if it takes effect as legislated, is expected to raise the delivered cost of hydrogen at a typical European site in 2032 by 20% compared with annual matching (median best estimate; IQR 16 to 35) (Q91, n=21). China's national ETS, which steel entered in 2025 with intensity-based free allocation, is expected to deliver a modest reduction in sector emissions by 2030, mainly through efficiency gains with little route switching, by 52%; 20% say it is too early to judge, 20% expect a negligible effect and 8% a material reduction (Q92, n=50).

Asked what requirements will apply by 2040 when an existing blast furnace seeks permission to reline, experts expect the European Union to prohibit relining of unabated furnaces (51%) or to require abatement (33%). For China the modal answer is an emissions-reduction plan (46%) with 33% expecting abatement; for Japan abatement (47%) or an emissions-reduction plan (40%); for Korea similar (43% and 39%, with 13% prohibition). For India (47% no specific climate requirement, 47% a plan) and the United States (36% and 40%, with 13% for abatement and 11% for prohibition) few expect abatement to be required (Q93, n=55).

All 11 questions in this module

Synthesis

Where experts agree and disagree

Dispersion scores each question from 0, where every expert agreed, to 1, where answers spread across the whole range. The ten most agreed and ten most contested questions are listed below.

Most agreed

Lowest dispersion · score 0 (unanimous) to 1

  1. Q50 For feed charged to a shaft furnace as DR pellets or lump ore, below what Fe grade of that feed can the plant… n=17 0.05
  2. Q69 What will global crude steel demand be in 2050, all routes and regions? n=30 0.15
  3. Q8 Allocate global crude steel production in 2040 across production routes (percent, summing to 100). n=88 0.17
  4. Q76 Allocate China's domestic crude steel production in 2035 across production routes (percent, summing to 100). n=28 0.18
  5. Q38 Allocate global hydrogen-based DRI/HBI production in 2040 across producing regions (percent, summing to 100). n=30 0.20
  6. Q82 And by 2050, what share of each region's primary iron units for near-zero steel will be imported as HBI or DR… n=32 0.23
  7. Q25 For a representative first-of-a-kind commercial-scale hydrogen shaft-furnace DRI plant, what capacity factor … n=33 0.23
  8. Q20 What total electricity input per tonne of crude steel will a best-in-class hydrogen-based DRI-EAF plant need … n=33 0.25
  9. Q86 How important is each safeguard for emissions-accounting schemes covering near-zero and low-emission steel? n=67 0.25
  10. Q24 Based on publicly available evidence, how important is each technical challenge for scaling hydrogen-based DR… n=32 0.27

Most contested

Highest dispersion · score 0 (unanimous) to 1

  1. Q13 What share of global hydrogen-based DRI/HBI output will be traded internationally as merchant iron in 2040, r… n=71 1.00
  2. Q44 How much hydrogen-based merchant iron (DRI/HBI) will be traded internationally in 2035? n=24 1.00
  3. Q29 What will total water consumption be for a best-in-class hydrogen-based DRI-EAF plant in 2035, including elec… n=16 1.00
  4. Q46 In 2030, what will be the gap between the DR-grade pellet demand of DRI plants then in commercial operation a… n=16 1.00
  5. Q87 By what year, if ever, will one steel product carbon-intensity standard, with shared definitions, system boun… n=62 1.00
  6. Q35 How will most hydrogen-ready DRI plants use their reductant flexibility in 2035? Rank the three most likely m… n=27 0.97
  7. Q58 By 2030, which contract structure will be most common for large-volume green iron and steel offtake agreement… n=15 0.97
  8. Q47 If global high-grade ore supply grew substantially by 2030 (say an additional 100 Mt per year of ore above 65… n=29 0.96
  9. Q12 Rank the three policy instruments that would most accelerate hydrogen-based ironmaking deployment over the ne… n=92 0.95
  10. Q91 If the EU's hourly renewable-matching requirement for RFNBO hydrogen takes effect as legislated, by what perc… n=21 0.95
Read the full analysis

Each question was scored for dispersion on a 0 to 1 scale: normalised entropy of the shares for choice and ranking questions, and the interquartile range relative to the scale (ratings, probabilities, allocations), to the median (quantities) or to ten years (year questions). The profile questions are excluded. Because the two families of questions are scored differently, the most agreed questions are mostly numeric and the most contested are mostly choices and rankings, so the two lists should be read within each family rather than against each other.

The ten most tightly clustered answers are: the shaft-furnace feed grade below which a plant cannot operate commercially, 65% Fe (Q50; dispersion 0.05); global crude steel demand in 2050, 2,025 Mt (Q69; 0.15); the global 2040 route mix (Q8; 0.17); China's 2035 route mix (Q76; 0.18); the regional allocation of 2040 hydrogen-based iron production (Q38; 0.20); 2050 import shares for near-zero steel by region (Q82; 0.23); the first-two-year capacity factor of a first-of-a-kind plant, 65% (Q25; 0.23); electricity input per tonne of crude steel, 4.0 MWh (Q20; 0.25); the importance of safeguards for emissions-accounting schemes (Q86; 0.25); and the importance of the technical challenges for scaling hydrogen-based DRI (Q24; 0.27). The ratings questions in this list agree because experts rate nearly everything as important; the numeric ones agree in substance. With the final 16 responses the importance of public investment in hub infrastructure (Q41) dropped out of the list, as the larger supply-chain group spread its ratings more widely, and the accounting safeguards took its place.

The ten most contested answers are: the share of hydrogen-based iron traded internationally in 2040 (Q13; dispersion 1.00), the volume of merchant iron traded in 2035 (Q44; 1.00), the water consumption of a best-in-class plant in 2035 (Q29; 1.00) and the 2030 DR-grade pellet gap (Q46; 1.00), all with interquartile ranges as wide as their medians, together with the year by which a single mutually recognised steel carbon-intensity standard will exist, whose middle half spans more than a decade (Q87; 1.00); how hydrogen-ready plants will use reductant flexibility (Q35; 0.97); the most common offtake contract structure in 2030 (Q58; 0.97); the effect of 100 Mt of extra high-grade ore on DR-grade availability (Q47; 0.96); the policy accelerators (Q12; 0.95); and the cost penalty of hourly renewable matching for RFNBO hydrogen (Q91; 0.95). The common thread is trade, inputs, commercial structure and the rules of the game: experts broadly agree on technology and on the shape of the route mix, and disagree on how much iron will move across borders, how much water and pellet feed the first plants will need, under what contracts projects will sell, and when the accounting rules will settle. The delivered cost of hydrogen in 2035 (Q62), among the most contested questions in the earlier cut, left the list as the final responses narrowed its interquartile range from USD 2.5 to 5.5 per kilogram to USD 2.5 to 4.4; the most useful Task 52 outputs (Q7) and the reline decision (Q73) sit just below the cut at 0.95 and 0.93.

Breakdowns by Task 52 affiliation

Subtask A experts put energy-price risk first among barriers to investment while every other group puts offtake first; the industry group is the most sure that new DRI projects will start on natural gas; and the technology group is the most willing to count upstream emissions in the definition of near-zero steel.

Subtask A
27
Subtask B
16
Subtask C
19
Linkage and exchange
12
Industry and policy reference
21
Not yet assigned
8

Per-question results were also computed for each Task 52 affiliation group with at least five answers (Subtask A, Subtask B, Subtask C, Task linkage and exchange, industry or policy reference experts, and not yet assigned). Most expertise-module questions only reach that threshold for Subtask A and the industry group, although the markets module now clears it for every group, and several of the differences below rest on groups of five to ten experts, so they are leads for discussion rather than findings.

On timing and cost the groups are close. The pooled 50% year for 10 Mt a year of hydrogen DRI capacity is 2036 for Subtask C (n=15), 2037 for Subtask A (n=25), Subtask B (n=16) and the linkage group (n=11) and 2040 for the industry group (n=15) (Q9). The 2040 route-mix medians differ by a few points at most, with the linkage group and Subtask B giving hydrogen DRI-EAF the highest median shares, 11% (n=11) and 10% (n=15), against 5 to 6% elsewhere (Q8). Subtask C is the most optimistic on the 2035 cost premium, with 62% of its members expecting 10 to 25% (n=16), and Subtask A on 2040, where half expect a 0 to 10% premium (n=22); the industry group still centres on 10 to 50% in 2040, with only 23% expecting 10% or less (n=13) (Q10).

Barriers and policy diverge by vantage point. Subtask A puts hydrogen or electricity price risk first as the barrier to final investment decision (71% of maximum Borda, n=26), while Subtask C (70%, n=18), the industry group (67%, n=20), Subtask B (60%, n=16) and the unassigned group (86%, n=7) put offtake and green-premium uncertainty first (Q11). As accelerators, Subtask B is split between border carbon adjustment and domestic carbon pricing (38% and 33%, n=15), Subtask A between capital grants and lead markets (38% each, n=22), while the industry group leans to lead markets (38%) over capital grants (33%, n=20); Subtask C (42%, n=16) and the linkage group (42%, n=12) put lead markets first, and the unassigned group hydrogen production tax credits (38%, n=7) (Q12). On the definition boundary, 74% of Subtask A want upstream Scope 3 included (n=23), against 50% of Subtask B (n=16) and 42% of the industry group (n=19), where Scope 1 and 2 only draws 38% and 37% (Q15). The industry group largely agrees that commercial DRI projects will start on natural gas (78%, n=18); Subtask B is split evenly between gas-first and a regional mix (47% each, n=15) (Q16).

Within the technology module, industry reference experts rate shaft-furnace DRI on 100% hydrogen less mature than Subtask A does (36% TRL 5 to 6, n=11, against 14% in Subtask A, n=22, with 9% and 14% at TRL 9) (Q17), yet expect the first 1 Mt plant sooner (20% already operating and 50% by 2028, n=10, whereas 55% of Subtask A say 2029 to 2032, n=20) (Q18). Subtask A ranks ore grade and feed flexibility as the most valuable form of flexibility (58%, n=16) while the industry group ranks reductant switching and locational flexibility jointly first (63% each, n=8) (Q36), and Subtask A expects seasonal variation of the hydrogen share to be the most common operating mode (57%, n=17) while the industry group expects plants to run mostly on gas with hydrogen when subsidised (67%, n=7) (Q35).

In the markets module, Subtask B (42%, n=12), Subtask A (41%, n=17) and the linkage group (60%, n=10) centre the 2030 premium on USD 50 to 100 per tonne, Subtask C leans towards up to USD 50 (56%, n=9), and the industry group is the most dispersed, leaning towards up to USD 50 (43%) with 14% expecting no premium and 14% more than USD 200 (n=14) (Q83). On CBAM's treatment of imported iron, 38% of Subtask B (n=8) and 33% of Subtask A (n=12) expect it to favour EU production, 67% of the industry group expect a neutral effect (n=15), 33% of the linkage group expect it to favour verified low-carbon imports (n=9) and 43% of Subtask C call it too uncertain (n=7) (Q89). Forty per cent of the industry group expects China's ETS to have a negligible effect on emissions by 2030 (n=10), whereas 64% of Subtask B (n=11) and 86% of the linkage group (n=7) expect a modest reduction (Q92). Subtask B ranks the EU CBAM extension to downstream products as the policy most likely to be weakened (48%, n=11) while every other group puts the EU ETS free-allocation phase-out first, most strongly the industry group (79%, n=13), the linkage group (78%, n=9) and Subtask C (70%, n=9) (Q88).

Preferences for Task 52 outputs follow the Subtask remits: 75% of Subtask B want the supply-chain route and cost comparison tool (n=16), 56% of Subtask A the technology readiness atlas (n=27), 58% of Subtask C the open techno-economic database and 47% the comparison of global scenario studies (n=19), and the industry group the scenario comparison (43%) and the policy brief series (38%, n=21) (Q7).

For the Task

What this means for Task 52

  1. 1

    Build the open techno-economic database and the readiness atlas around realised capex, operating cost, availability and capacity-factor evidence from demonstration and first commercial plants: experts chose the database first (57%; Q7, n=104) and find exactly these data hardest to obtain Q23, n=39.

  2. 2

    Keep merchant iron and integrated on-site steelmaking apart in every cost comparison; experts regard their costs and risks as different.

  3. 3

    Concentrate the markets and policy work on accounting safeguards and mutual recognition, the gap that most limits decisions today Q85, n=63 and one not expected to close before about 2040 Q87, n=62.

  4. 4

    Test the network’s central expectations, 6% hydrogen DRI and 42% unabated BF-BOF in 2040, against published scenarios in the Subtask C comparison and explain any gap.

  5. 5

    Validate the finance results with lenders and project developers: that module drew 26 experts and nobody from finance or investment.

  6. 6

    Run a second, structured round with feedback on the most contested topics: trade volumes, the DR-grade pellet gap, plant water use and the timing of a carbon-intensity standard.

Read the full text

The outputs experts asked for match the gaps they identified. An open techno-economic database of routes was chosen by 57% of all respondents, a comparison of global scenario studies and a supply-chain route and cost comparison tool by 38% each, a technology readiness atlas by 37% and a global project tracker by 36% (Q7, n=104). The data experts find hardest to obtain are operating costs and availability of demonstration plants (77%) and realised capex of first plants (69%) (Q23), and the knowledge gap that most limits decisions is carbon accounting and certification rules (Q85). Three actions follow. First, the techno-economic database and readiness atlas should prioritise realised capex, operating cost, availability and capacity-factor evidence from demonstration and first commercial plants, including the 65% first-two-year capacity factor and 4.0 MWh per tonne benchmarks elicited here, and should separate merchant iron from integrated on-site steelmaking, whose costs and risks experts regard as different. Second, the markets and policy work should concentrate on accounting safeguards and mutual recognition, where every safeguard was rated above 4 of 5 and a recognised standard is not expected before about 2040. Third, the Subtask C scenario comparison should test the network's central expectations, a 6% hydrogen DRI share and 42% unabated BF-BOF in 2040, against published pathways and explain the gap.

The elicitation also tells the Task where its own evidence is thin. The finance module drew 26 experts and no respondent came from finance or investment, so the cost-of-debt and risk-allocation results need validation with lenders and project developers. Nobody named Sub-Saharan Africa as a region of expertise, and Türkiye, the Middle East and North Africa were raised as regional pathways to add. The most contested results, on trade volumes, the DR-grade pellet gap, plant water consumption and the timing of a single carbon-intensity standard, are where a second, structured round with feedback would most reduce uncertainty; the delivered cost of hydrogen, equally contested in the earlier cut, tightened as the sample grew, which suggests that some of the remaining spread is small-sample noise rather than settled disagreement.

For the Oxford panels and the COP31 launch, the results give each panel a quantified starting point: technology readiness and process integration (Q17 to Q29), global value chains, bankability and resources (Q38 to Q59), enabling projects through supply chains, infrastructure and policy (Q60 to Q75) and enabling markets through accounting, procurement and standards (Q83 to Q93). The free-text questions experts submitted for those panels have been summarised separately for the Task leadership.

How the elicitation was run and its limits

Task 52 (Hydrogen for Iron and Steelmaking) is a Task of the IEA Hydrogen Technology Collaboration Programme. Its expert network spans technology assessment (Subtask A), resources, supply chains and infrastructure (Subtask B), global pathways, competitiveness and power systems (Subtask C), a linkage group that contributes across Subtasks, and industry and policy reference experts.

The elicitation was a structured online questionnaire of 93 questions in nine modules, built from the question bank version 2026-09-oxford-v4 after two rounds of review by the Task leadership. Two modules were core and answered by everyone: a profile module (7 questions, including which Task 52 outputs the expert would find most useful) and a cross-cutting outlook module (9 questions on the 2040 route mix, timing, costs, barriers, policy and definitions). Seven expertise modules were self-selected: technology readiness and process integration (13 questions), process flexibility (8), resources, supply chains and infrastructure (16), projects, finance and offtake (6), pathways, competitiveness and power systems (16), regional pathways (7) and markets, standards and policy (11).

Question formats were chosen to make uncertainty explicit. Quantities were asked as a low, high and best estimate; years as the year by which the expert sees a 10%, 50% and 90% chance; preferences as ranked top-three lists, point allocations summing to 100, probability sliders, choice grids and five-point ratings. A one-minute calibration warm-up on two known quantities (world crude steel production in 2024 and world DRI production in 2023) preceded the modules. Experts could save a draft under a resume code and return later.

The questionnaire opened on 12 September 2026 and was circulated to the Task 52 expert network ahead of the Oxford workshop on the economic competitiveness of hydrogen-based iron and steelmaking (University of Oxford, 25 September 2026). The nominal closing date was 24 September 2026; the questionnaire was then reopened until Friday 30 October 2026 so that the pooled results can feed the Task 52 annual report, and a few responses that arrived in the days after the closing date were accepted. This report uses all responses received up to 8 November 2026. It was a single round with no group feedback: experts did not see each other's answers, so the spread in every result is genuine disagreement rather than a converged consensus. This report covers 104 responses, one per expert (the latest submission per email address), with data as of 8 November 2026. The pooled results from the first round (74 responses) were presented at the Oxford workshop, where they framed the four panel discussions; an interim cut of 84 responses was used to prepare this report, and the full set of 104 will be launched publicly at COP31.

The expert base in detail

One hundred and four experts submitted a response. Of the 103 who stated an affiliation within the Task, 26% are most closely associated with Subtask A (technology assessment and readiness), 16% with Subtask B (resources, supply chains and infrastructure), 18% with Subtask C (global pathways, competitiveness and power systems), 12% with Task linkage and exchange, 20% are industry or policy reference experts and 8% are not yet assigned (Q1, n=103).

By sector, universities and research institutes supplied 40% of respondents, NGOs, think tanks and industry associations 15%, technology providers, EPC and equipment companies 10%, intergovernmental organisations 10%, iron and steel producers 8%, consultancies 7%, mining and ore suppliers 4%, energy, hydrogen and utility companies 4% and government agencies 4%. The latest responses came from intergovernmental organisations, an NGO or think tank and a government agency; no respondent came from finance or investment (Q3, n=103).

Experts could name up to three regions of expertise. Europe was named by 49%, a global or non-specific focus by 31%, Australia and New Zealand by 23%, China by 21%, India and South-East Asia by 14%, Japan and South Korea by 11%, the Middle East and North Africa by 10%, Brazil and Latin America by 9% and North America by 7%. Nobody named Sub-Saharan Africa (Q4, n=103). Australian expertise is better represented than in the earlier cut, reflecting where a large share of the final-week responses came from. Professional backgrounds (up to two each) were led by economists and techno-economic modellers (32%), policy or regulatory analysts (26%), metallurgists and materials scientists (24%), chemical and process engineers (19%) and business, strategy and finance professionals (17%) (Q2, n=103). Experience is split between newer and long-standing specialists: 36% of respondents have worked on iron and steel for 3 to 6 years and 20% for less than 3 years, while 28% have more than 15 years and 16% have 7 to 15 years (Q5, n=102).

Self-rated expertise on a five-point scale was highest for ironmaking and steelmaking process technology (mean 3.5, median 4), followed by techno-economics, markets and trade (3.4, median 3), renewable electricity, hydrogen production and power systems (3.3, median 3), policy, standards, certification and finance (3.1), supply chains, logistics and infrastructure (2.8) and iron ore quality, beneficiation and pelletising (2.7) (Q6, n=102). The two calibration items were answered by 97 and 93 experts; the stated low-to-high ranges contained the true value 71% of the time for world crude steel production and 63% of the time for world DRI production, better than the roughly 50% hit rate the warm-up warned of, but below the 80% a well-calibrated range should reach.

Module uptake reflects the network's centre of gravity. Beyond the two core modules (104 each), markets, standards and policy drew 81 experts, technology readiness 47, regional pathways 42, pathways and power systems 42, supply chains 37, process flexibility 34 and projects and finance 26. The median expert answered five modules including the two core ones, and 36 questions (IQR 25 to 50; range 7 to 93); seven experts answered all nine modules. The median time between starting and submitting was 62 minutes, with a quarter finishing in under 41 minutes and a long tail of sessions saved and resumed over days or weeks. Asked at the end how confident they were overall, 78 experts rated their 2030 answers (65% moderate, 28% high, 6% low) and 65 rated their 2035 to 2040 answers (65% moderate, 31% low, 5% high). Thirty-two experts left 87 free-text comments for the Oxford panels; these were used to shape the panel discussions and are not reproduced.

Methodology and limitations

Central values are medians and spread is the interquartile range, the middle half of the answers; charts and tables also show the full range where it is informative. For four-value questions the statistics are of the best estimates, with the medians of the low and high bounds reported alongside. Choice questions report the share of experts choosing each option; multiple-choice questions report the share selecting each option, so shares sum to more than 100%. Ranking questions use a Borda count in which a first place scores three points, second two and third one; the Borda share is the score as a fraction of the maximum possible, and the share ranking an option first and the share mentioning it anywhere in the top three are also reported. Allocation questions report the median and interquartile range of the points given to each option, so medians need not sum to 100. Ratings report the mean, median and interquartile range of the one-to-five scores, with rows marked not my expertise omitted. Year questions pool cumulative curves: each expert gives the year by which they see a 10%, 50% and 90% chance, and the pooled curve is the average of their piecewise-linear cumulative distributions, with never treated as beyond the horizon. Agreement is scored by normalised entropy for choices and rankings and by relative interquartile range for numeric, probability, allocation and year questions.

The sample is the Task 52 expert network, not a random sample of the industry. Expertise modules were self-selected, so each result rests on a different and often small group: 26 experts in finance, 34 in flexibility, 37 in supply chains, 42 in pathways, 42 in regional pathways, 47 in technology and 81 in markets, and individual questions within those modules can rest on fewer than a dozen answers where experts skipped a question or marked it outside their expertise. Every result in this report and its appendix states its n. Academia supplied two in five respondents, Europe was the most cited region of expertise and no respondent came from finance. It was a single round with no group feedback, so the results show where independent judgements fall, not a negotiated consensus. The calibration warm-up found that stated ranges contained the true value 71% and 63% of the time on two known quantities, so the low-to-high ranges in the four-value questions are probably somewhat narrower than they should be. Estimates are judgements about an uncertain future made between September and early November 2026, not measurements.

The questionnaire closed on 30 October 2026. This report covers 104 responses, one per expert (the latest submission per email address), with data as of 8 November 2026; a small number of responses that arrived in the days after the closing date were accepted and are included. Anonymity rules were set before the elicitation opened. Names, organisations, email addresses and free-text comments never enter the published dataset; only aggregates are released. Breakdowns by Task 52 affiliation are shown only where a group has at least five answers to the question, and no other breakdowns are published. Experts were told their responses were confidential rather than anonymous, and that pooled findings would be shared with the network and the public.

Question bank 2026-09-oxford-v4. Data as of 8 November 2026. No individual expert, organisation or free-text comment is identifiable; breakdowns by Task 52 affiliation are shown only for groups of five or more.

Global Steel Transition Atlas

The map behind the expectations

The Atlas was launched alongside this report. It is an open, interactive map of where steel is made, where the ore is, how iron moves, where the emissions are, where green hydrogen will be cheapest and where the green projects are. Five views are shown here as they stood on 11 November 2026; the figures move as the underlying trackers update.

Open the Atlas

Where steel is made and where the ore is
1

Where steel is made and where the ore is

  • 1,293 steel plants tracked; 2.2 Bt a year of operating crude steel capacity
  • 980 iron ore mines, 701 operating; 4.06 Gt a year of operating design capacity
  • Australia leads operating ore capacity (1,208 Mt a year), then Brazil (605) and India (506)

Steel plants by route (EAF, BF-BOF, DRI) and iron ore mines, sized by capacity. Data: Global Energy Monitor, Global Iron & Steel Tracker and Global Iron Ore Mines Tracker, September 2026 (CC BY 4.0).

How iron ore moves today: Australia and Brazil to China
2

How iron ore moves today: Australia and Brazil to China

  • 1,698 Mt of iron ore traded in 2024 across 1,040 bilateral links
  • Australia is the top exporter (893 Mt), then Brazil (401 Mt)
  • China is the top importer (1,246 Mt); the Atlas also shows DRI/HBI, pig iron, scrap and steel flows

Iron ore exporter-to-importer flows, width by tonnage. Data: CEPII BACI international trade database, HS22, 2024 (release 202601).

Emissions: 3.4 Gt a year, mostly from coal blast furnaces
3

Emissions: 3.4 Gt a year, mostly from coal blast furnaces

  • 3.36 Gt CO2 a year modelled for the 970 operating plants covered (range 3.1 to 3.9)
  • 83% comes from the BF-BOF coal route; implied intensity 1.87 t CO2 per tonne of crude steel
  • 134 Mt a year is avoidable if the announced green pipeline displaces BF-BOF output

Estimated CO2 by plant, colour by intensity, size by emissions; operating plants only. Modelled with route emission factors applied to reported production or capacity; grid intensity from Our World in Data / Ember, 2024. A model, not measurements.

Where green hydrogen will be cheapest in 2030
4

Where green hydrogen will be cheapest in 2030

  • Levelised cost of green hydrogen in 2030 on a 1-degree grid; the colour ramp runs from 2 to 13 USD per kg
  • Hourly least-cost optimisation of wind, solar, batteries, electrolysers and storage for every onshore cell, weather year 2023
  • Modelled estimates, not market prices; 2040 and 2050 surfaces, electricity and ammonia are also in the Atlas

Green hydrogen cost surface, 2030, pilot 1-degree grid; blank cells were not modelled or were screened out. Data: Global Green Cost Maps, Monash University (Changlong Wang and colleagues), 2026.

Green projects: 180 announced, 104 hydrogen-based
5

Green projects: 180 announced, 104 hydrogen-based

  • 180 green-steel projects with 86 Mt a year of announced capacity and USD 98.7 billion of disclosed investment
  • 104 are hydrogen-based; the rest are gas DRI, CCUS, scrap EAF, biomass or novel routes
  • 129 hydrogen-ready DRI furnaces built or announced, 52 operating, 7 running on hydrogen today

Green projects and CCS, sized by announced capacity, linked to host plants. Data: LeadIT Green Steel Tracker, 2026 Q1 (158 projects) and CISRI / MCC China hydrogen-metallurgy projects, 2026 (22), matched to Global Energy Monitor plants.

All results

Every question, with its chart and its numbers

Central values are medians and spread is the interquartile range; n is the number of experts who answered. Open a module to see its questions.

About you 104 experts · 7 questions
Q1 Single choice n = 103

Which part of Task 52 are you most closely associated with?

Subtask A — Technology assessment and readiness: 26% (27) Subtask A — Technology assessment and readine… 26% (27) Subtask B — Resources, supply chains and infrastructure: 16% (16) Subtask B — Resources, supply chains and infr… 16% (16) Subtask C — Global pathways, competitiveness and power systems: 18% (19) Subtask C — Global pathways, competitiveness … 18% (19) Task linkage and exchange — contributing across more than one Subtask, policy advice and international exchange: 12% (12) Task linkage and exchange — contributing acro… 12% (12) Industry or policy reference expert: 20% (21) Industry or policy reference expert 20% (21) Other / not yet assigned: 8% (8) Other / not yet assigned 8% (8)
n = 103
OptionExpertsShare
Subtask A — Technology assessment and readiness2726%
Subtask B — Resources, supply chains and infrastructure1616%
Subtask C — Global pathways, competitiveness and power systems1918%
Task linkage and exchange — contributing across more than one Subtask, policy advice and international exchange1212%
Industry or policy reference expert2120%
Other / not yet assigned88%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=27Subtask A — Technology assessment and readiness 100%
  • Subtask Bn=16Subtask B — Resources, supply chains and infrastructure 100%
  • Subtask Cn=19Subtask C — Global pathways, competitiveness and power systems 100%
  • Linkage and exchangen=12Task linkage and exchange — contributing across more than one Subtask, policy advice and international exchange 100%
  • Industry and policy referencen=21Industry or policy reference expert 100%
  • Not yet assignedn=8Other / not yet assigned 100%
Q2 Multiple choice, up to 2 n = 103

What is your professional background? Select up to two.

Economist or techno-economic modeller: 32% Economist or techno-economic modeller 32% Policy or regulatory analyst: 26% Policy or regulatory analyst 26% Metallurgist or materials scientist: 24% Metallurgist or materials scientist 24% Chemical or process engineer: 19% Chemical or process engineer 19% Business, strategy or finance professional: 17% Business, strategy or finance professional 17% Mechanical, plant or project engineer: 11% Mechanical, plant or project engineer 11% Electrical, energy-systems or power engineer: 9% Electrical, energy-systems or power engineer 9% Other: 9% Other 9% Physical or environmental scientist: 7% Physical or environmental scientist 7% Mining engineer or geologist: 2% Mining engineer or geologist 2%

Share of respondents selecting each option.

n = 103 · each expert could select up to 2
OptionExpertsShare selecting
Economist or techno-economic modeller3332%
Policy or regulatory analyst2726%
Metallurgist or materials scientist2524%
Chemical or process engineer2019%
Business, strategy or finance professional1817%
Mechanical, plant or project engineer1111%
Electrical, energy-systems or power engineer99%
Other99%
Physical or environmental scientist77%
Mining engineer or geologist22%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=27Metallurgist or materials scientist 59%
  • Subtask Bn=16Economist or techno-economic modeller 31%
  • Subtask Cn=19Economist or techno-economic modeller 68%
  • Linkage and exchangen=12Economist or techno-economic modeller 58%
  • Industry and policy referencen=21Policy or regulatory analyst 52%
  • Not yet assignedn=8Other 38%
Q3 Single choice n = 103

Which sector best describes your current organisation?

Industry — iron or steel producer: 8% (8) Industry — iron or steel producer 8% (8) Industry — mining or ore supplier: 4% (4) Industry — mining or ore supplier 4% (4) Industry — technology provider, EPC or equipment: 10% (10) Industry — technology provider, EPC or equipm… 10% (10) Industry — energy, hydrogen or utility: 4% (4) Industry — energy, hydrogen or utility 4% (4) Finance or investment: 0% (0) Finance or investment 0% (0) Government or public agency: 4% (4) Government or public agency 4% (4) Intergovernmental or international organisation: 10% (10) Intergovernmental or international organisati… 10% (10) University or research institute: 40% (41) University or research institute 40% (41) NGO, think tank or industry association: 15% (15) NGO, think tank or industry association 15% (15) Consultancy: 7% (7) Consultancy 7% (7)
n = 103
OptionExpertsShare
Industry — iron or steel producer88%
Industry — mining or ore supplier44%
Industry — technology provider, EPC or equipment1010%
Industry — energy, hydrogen or utility44%
Finance or investment00%
Government or public agency44%
Intergovernmental or international organisation1010%
University or research institute4140%
NGO, think tank or industry association1515%
Consultancy77%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=27University or research institute 41%
  • Subtask Bn=16University or research institute 56%
  • Subtask Cn=19University or research institute 63%
  • Linkage and exchangen=12University or research institute 42%
  • Industry and policy referencen=21NGO, think tank or industry association 24%
  • Not yet assignedn=8Intergovernmental or international organisation 25%
Q4 Multiple choice, up to 3 n = 103

Which regions does your expertise mainly relate to? Select up to three.

Europe: 49% Europe 49% Global / no specific region: 31% Global / no specific region 31% Australia and New Zealand: 23% Australia and New Zealand 23% China: 21% China 21% India and South-East Asia: 14% India and South-East Asia 14% Japan and South Korea: 11% Japan and South Korea 11% Middle East and North Africa: 10% Middle East and North Africa 10% Brazil and Latin America: 9% Brazil and Latin America 9% North America: 7% North America 7% Sub-Saharan Africa: 0% Sub-Saharan Africa 0%

Share of respondents selecting each option.

n = 103 · each expert could select up to 3
OptionExpertsShare selecting
Europe5049%
Global / no specific region3231%
Australia and New Zealand2423%
China2221%
India and South-East Asia1414%
Japan and South Korea1111%
Middle East and North Africa1010%
Brazil and Latin America99%
North America77%
Sub-Saharan Africa00%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=27Europe 63%
  • Subtask Bn=16China 44%
  • Subtask Cn=19Europe 74%
  • Linkage and exchangen=12Europe 50%
  • Industry and policy referencen=21Australia and New Zealand 29%
  • Not yet assignedn=8Global / no specific region 50%
Q5 Single choice n = 102

How many years have you worked professionally on iron and steel or closely related fields?

Less than 3 years: 20% (20) Less than 3 years 20% (20) 3 to 6 years: 36% (37) 3 to 6 years 36% (37) 7 to 15 years: 16% (16) 7 to 15 years 16% (16) More than 15 years: 28% (29) More than 15 years 28% (29)

Mean score 9.3, median 4.5

n = 102 · mean score 9.34, median 4.5
OptionExpertsShare
Less than 3 years2020%
3 to 6 years3736%
7 to 15 years1616%
More than 15 years2928%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=26More than 15 years 58%
  • Subtask Bn=163 to 6 years 38%
  • Subtask Cn=193 to 6 years 58%
  • Linkage and exchangen=123 to 6 years 67%
  • Industry and policy referencen=21More than 15 years 29%
  • Not yet assignedn=8Less than 3 years 38%
Q6 Rating, 1 (General awareness) to 5 (Leading expert) n = 102

Rate your own expertise in each area.

1 = general awareness, 3 = working knowledge, 5 = leading expert. Used to weight answers within modules, never to exclude them.

Ironmaking and steelmaking process tech… Ironmaking and steelmaking process technology · 1: 4% (n=96)Ironmaking and steelmaking process technology · 2: 15% (n=96)Ironmaking and steelmaking process technology · 3: 29% (n=96)Ironmaking and steelmaking process technology · 4: 31% (n=96)Ironmaking and steelmaking process technology · 5: 21% (n=96) Mean score (n=96)3.5 Techno-economics, markets and trade Techno-economics, markets and trade · 1: 7% (n=97)Techno-economics, markets and trade · 2: 15% (n=97)Techno-economics, markets and trade · 3: 28% (n=97)Techno-economics, markets and trade · 4: 32% (n=97)Techno-economics, markets and trade · 5: 18% (n=97) Mean score (n=97)3.4 Renewable electricity, hydrogen product… Renewable electricity, hydrogen production and power systems · 1: 4% (n=98)Renewable electricity, hydrogen production and power systems · 2: 10% (n=98)Renewable electricity, hydrogen production and power systems · 3: 48% (n=98)Renewable electricity, hydrogen production and power systems · 4: 22% (n=98)Renewable electricity, hydrogen production and power systems · 5: 15% (n=98) Mean score (n=98)3.3 Policy, standards, certification and fi… Policy, standards, certification and finance · 1: 9% (n=95)Policy, standards, certification and finance · 2: 22% (n=95)Policy, standards, certification and finance · 3: 35% (n=95)Policy, standards, certification and finance · 4: 19% (n=95)Policy, standards, certification and finance · 5: 15% (n=95) Mean score (n=95)3.1 Supply chains, logistics and infrastruc… Supply chains, logistics and infrastructure · 1: 10% (n=92)Supply chains, logistics and infrastructure · 2: 27% (n=92)Supply chains, logistics and infrastructure · 3: 46% (n=92)Supply chains, logistics and infrastructure · 4: 12% (n=92)Supply chains, logistics and infrastructure · 5: 5% (n=92) Mean score (n=92)2.8 Iron ore quality, beneficiation and pel… Iron ore quality, beneficiation and pelletising · 1: 16% (n=91)Iron ore quality, beneficiation and pelletising · 2: 24% (n=91)Iron ore quality, beneficiation and pelletising · 3: 37% (n=91)Iron ore quality, beneficiation and pelletising · 4: 14% (n=91)Iron ore quality, beneficiation and pelletising · 5: 8% (n=91) Mean score (n=91)2.7
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Ironmaking and steelmaking process technology963.504.03.0 to 4.0
Techno-economics, markets and trade973.373.03.0 to 4.0
Renewable electricity, hydrogen production and power systems983.353.03.0 to 4.0
Policy, standards, certification and finance953.073.02.0 to 4.0
Supply chains, logistics and infrastructure922.763.02.0 to 3.0
Iron ore quality, beneficiation and pelletising912.733.02.0 to 3.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=26Ironmaking and steelmaking process technology 4.2
  • Subtask Bn=16Techno-economics, markets and trade 3.8
  • Subtask Cn=19Techno-economics, markets and trade 3.6
  • Linkage and exchangen=12Policy, standards, certification and finance 3.7
  • Industry and policy referencen=21Ironmaking and steelmaking process technology 3.6
  • Not yet assignedn=8Renewable electricity, hydrogen production and power systems 3.4
Q7 Multiple choice, up to 3 n = 104

Which Task 52 outputs would be most useful to you and your organisation? Select up to three.

Everyone answers this one: it shapes what Task 52 publishes first.

Open techno-economic database of routes: 57% Open techno-economic database of routes 57% Supply-chain route and cost comparison tool: 38% Supply-chain route and cost comparison tool 38% Comparison of global scenario studies: 38% Comparison of global scenario studies 38% Technology readiness atlas: 37% Technology readiness atlas 37% Global project tracker: 36% Global project tracker 36% Workshop and expert dialogue series: 25% Workshop and expert dialogue series 25% Policy brief series: 20% Policy brief series 20% Joint peer-reviewed papers: 17% Joint peer-reviewed papers 17% Input to standards harmonisation: 12% Input to standards harmonisation 12% Researcher and staff exchange: 10% Researcher and staff exchange 10%

Share of respondents selecting each option.

n = 104 · each expert could select up to 3
OptionExpertsShare selecting
Open techno-economic database of routes5957%
Supply-chain route and cost comparison tool4038%
Comparison of global scenario studies4038%
Technology readiness atlas3837%
Global project tracker3736%
Workshop and expert dialogue series2625%
Policy brief series2120%
Joint peer-reviewed papers1817%
Input to standards harmonisation1212%
Researcher and staff exchange1010%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=27Technology readiness atlas 56%
  • Subtask Bn=16Supply-chain route and cost comparison tool 75%
  • Subtask Cn=19Open techno-economic database of routes 58%
  • Linkage and exchangen=12Open techno-economic database of routes 58%
  • Industry and policy referencen=21Open techno-economic database of routes 43%
  • Not yet assignedn=8Open techno-economic database of routes 75%
Cross-cutting outlook 104 experts · 9 questions
Q8 Allocation of 100 points n = 88

Allocate global crude steel production in 2040 across production routes (percent, summing to 100).

For reference, 2024 shares were roughly: BF-BOF 70%, scrap-based EAF 22%, natural-gas DRI-EAF 7%, other 1% (worldsteel). Enter your expected 2040 shares.

0%20%40%60%80%100% BF-BOF, unabated: median 42.0%, IQR 30.0–50.0, range 5.0–71.0 (n=88) BF-BOF, unabated Reference: 70% 42 BF-BOF with CCUS and/or hydrogen injection: median 5.0%, IQR 2.8–10.0, range 0.0–32.0 (n=88) BF-BOF with CCUS and/or hydrogen in… Reference: 0% 5 Natural-gas DRI-EAF, with or without CCS: median 12.0%, IQR 10.0–15.0, range 4.0–30.0 (n=87) Natural-gas DRI-EAF, with or withou… Reference: 7% 12 Hydrogen-based DRI-EAF, including imported green iron: median 5.6%, IQR 3.0–10.0, range 0.0–30.0 (n=88) Hydrogen-based DRI-EAF, including i… Reference: 0% 6 Scrap-based EAF: median 30.0%, IQR 25.0–35.0, range 10.0–60.0 (n=88) Scrap-based EAF Reference: 22% 30 Other novel routes (electrolysis, smelting reduction, plasma): median 2.0%, IQR 1.0–3.0, range 0.0–10.0 (n=79) Other novel routes (electrolysis, s… Reference: 1% 2

Box: interquartile range · thick line: median · whiskers: min–max · number: median · purple line: reference value

Points out of 100 · reference values are the baseline shown to experts, where one was given
OptionnMedianIQRRangeReference
BF-BOF, unabated8842.030.0 to 50.05 to 7170
BF-BOF with CCUS and/or hydrogen injection885.02.8 to 10.00 to 320
Natural-gas DRI-EAF, with or without CCS8712.010.0 to 15.04 to 307
Hydrogen-based DRI-EAF, including imported green iron885.63.0 to 10.00 to 300
Scrap-based EAF8830.025.0 to 35.010 to 6022
Other novel routes (electrolysis, smelting reduction, plasma)792.01.0 to 3.00 to 101

By Task 52 affiliation (groups of at least 5)

  • Subtask An=26BF-BOF, unabated 48
  • Subtask Bn=15BF-BOF, unabated 39
  • Subtask Cn=16BF-BOF, unabated 40
  • Linkage and exchangen=11BF-BOF, unabated 40
  • Industry and policy referencen=14BF-BOF, unabated 45
  • Not yet assignedn=6BF-BOF, unabated 41
Q9 Year percentiles n = 89

In which year will at least 10 million tonnes per year of hydrogen-based DRI capacity be in commercial operation globally?

Give the year by which you assign a 10% chance, a 50% chance and a 90% chance. Fill the 10% and 90% years first. Hydrogen-based means at least 90% hydrogen; today roughly 1 to 2 Mt/yr is under commissioning.

10% chance by

2032

IQR 2030–2035 · never: 1

50% chance by

2037

IQR 2035–2040 · never: 1

90% chance by

2042

IQR 2040–2050 · never: 2

0%25%50%75%100%20272031203520392043204720502027: 3% pooled probability2028: 4% pooled probability2029: 6% pooled probability2030: 10% pooled probability2031: 15% pooled probability2032: 20% pooled probability2033: 26% pooled probability2034: 32% pooled probability2035: 38% pooled probability2036: 45% pooled probability2037: 50% pooled probability2038: 56% pooled probability2039: 61% pooled probability2040: 67% pooled probability2041: 70% pooled probability2042: 74% pooled probability2043: 77% pooled probability2044: 79% pooled probability2045: 82% pooled probability2046: 84% pooled probability2047: 86% pooled probability2048: 87% pooled probability2049: 89% pooled probability2050: 90% pooled probability

Pooled cumulative probability: average of each expert's piecewise-linear curve through their 10%, 50% and 90% years.

“After 2050” is counted as 2055 for the statistics
PercentilenMedian yearIQRRange“Never”
10% chance by8820322030 to 20352027 to 20501
50% chance by8820372035 to 20402028 to 20551
90% chance by8720422040 to 20502030 to 20552

By Task 52 affiliation (groups of at least 5)

  • Subtask An=2550% chance by 2035
  • Subtask Bn=1650% chance by 2039
  • Subtask Cn=1550% chance by 2035
  • Linkage and exchangen=1150% chance by 2036
  • Industry and policy referencen=1650% chance by 2040
  • Not yet assignedn=650% chance by 2039
Q10 Choice grid n = 84

What production-cost premium will hydrogen-based DRI-EAF crude steel carry over BF-BOF in the best-in-class location, excluding carbon pricing and subsidies?

2030 2030 · Below BF-BOF cost: 1% (n=84)2030 · 0 to 10%: 6% (n=84)2030 · 10 to 25%: 19% (n=84)2030 · 25 to 50%: 46% (n=84)2030 · Above 50%: 27% (n=84) Mean score (n=84)37.4 2035 2035 · Below BF-BOF cost: 1% (n=83)2035 · 0 to 10%: 13% (n=83)2035 · 10 to 25%: 42% (n=83)2035 · 25 to 50%: 35% (n=83)2035 · Above 50%: 8% (n=83) Mean score (n=83)26.1 2040 2040 · Below BF-BOF cost: 5% (n=82)2040 · 0 to 10%: 37% (n=82)2040 · 10 to 25%: 40% (n=82)2040 · 25 to 50%: 13% (n=82)2040 · Above 50%: 5% (n=82) Mean score (n=82)16.6
Below BF-BOF cost0 to 10%10 to 25%25 to 50%Above 50%Number at right: mean score
Share of the experts answering each row
RownBelow BF-BOF cost0 to 10%10 to 25%25 to 50%Above 50%
2030841%6%19%46%27%
2035831%13%42%35%8%
2040825%37%40%13%5%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=222030: 25 to 50% 41%
  • Subtask Bn=152030: 25 to 50% 60%
  • Subtask Cn=162030: 25 to 50% 63%
  • Linkage and exchangen=112030: 10 to 25% 36%
  • Industry and policy referencen=152030: 25 to 50% 40%
  • Not yet assignedn=52030: Above 50% 80%
Q11 Ranked top 3 n = 98

Rank the three biggest barriers to reaching final investment decision on commercial-scale hydrogen-based ironmaking projects today.

Offtake and green-premium uncertainty: 63% · 1st 44% Offtake and green-premium uncertainty 63% · 1st 44% Hydrogen or electricity price risk: 53% · 1st 32% Hydrogen or electricity price risk 53% · 1st 32% Durability and predictability of policy support: 20% · 1st 4% Durability and predictability of policy suppo… 20% · 1st 4% Carbon price and CBAM implementation uncertainty: 18% · 1st 7% Carbon price and CBAM implementation uncertai… 18% · 1st 7% Cost of capital and balance-sheet constraints: 13% · 1st 4% Cost of capital and balance-sheet constraints 13% · 1st 4% First-of-a-kind technology risk: 9% · 1st 3% First-of-a-kind technology risk 9% · 1st 3% Capex escalation and EPC capacity: 8% · 1st 3% Capex escalation and EPC capacity 8% · 1st 3% Permitting and grid-connection delays: 7% · 1st 1% Permitting and grid-connection delays 7% · 1st 1% Security of DR-grade ore supply: 7% · 1st 1% Security of DR-grade ore supply 7% · 1st 1% Skilled workforce availability: 2% · 1st 1% Skilled workforce availability 2% · 1st 1%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 98 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Offtake and green-premium uncertainty63%44%80%
Hydrogen or electricity price risk53%32%70%
Durability and predictability of policy support20%4%38%
Carbon price and CBAM implementation uncertainty18%7%31%
Cost of capital and balance-sheet constraints13%4%23%
First-of-a-kind technology risk9%3%13%
Capex escalation and EPC capacity8%3%13%
Permitting and grid-connection delays7%1%13%
Security of DR-grade ore supply7%1%15%
Skilled workforce availability2%1%3%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=26Hydrogen or electricity price risk (Borda 71%)
  • Subtask Bn=16Offtake and green-premium uncertainty (Borda 60%)
  • Subtask Cn=18Offtake and green-premium uncertainty (Borda 70%)
  • Linkage and exchangen=11Offtake and green-premium uncertainty (Borda 58%)
  • Industry and policy referencen=20Offtake and green-premium uncertainty (Borda 67%)
  • Not yet assignedn=7Offtake and green-premium uncertainty (Borda 86%)
Q12 Ranked top 3 n = 92

Rank the three policy instruments that would most accelerate hydrogen-based ironmaking deployment over the next five years.

Lead markets or quotas for near-zero steel: 33% · 1st 16% Lead markets or quotas for near-zero steel 33% · 1st 16% Capital grants for first plants: 28% · 1st 17% Capital grants for first plants 28% · 1st 17% Carbon contracts for difference or production subsidies: 25% · 1st 14% Carbon contracts for difference or production… 25% · 1st 14% Domestic carbon pricing: 23% · 1st 17% Domestic carbon pricing 23% · 1st 17% Hydrogen production tax credits or subsidies: 21% · 1st 9% Hydrogen production tax credits or subsidies 21% · 1st 9% Green public procurement: 17% · 1st 4% Green public procurement 17% · 1st 4% Border carbon adjustment (CBAM): 16% · 1st 8% Border carbon adjustment (CBAM) 16% · 1st 8% Bilateral green iron and steel trade agreements: 16% · 1st 7% Bilateral green iron and steel trade agreemen… 16% · 1st 7% Product standards and embodied-carbon limits: 8% · 1st 3% Product standards and embodied-carbon limits 8% · 1st 3% Concessional finance, export credit or blended finance: 8% · 1st 3% Concessional finance, export credit or blende… 8% · 1st 3% Harmonised definitions and certification: 4% · 1st 1% Harmonised definitions and certification 4% · 1st 1%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 92 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Lead markets or quotas for near-zero steel33%16%47%
Capital grants for first plants28%17%37%
Carbon contracts for difference or production subsidies25%14%37%
Domestic carbon pricing23%17%29%
Hydrogen production tax credits or subsidies21%9%36%
Green public procurement17%4%27%
Border carbon adjustment (CBAM)16%8%24%
Bilateral green iron and steel trade agreements16%7%27%
Product standards and embodied-carbon limits8%3%14%
Concessional finance, export credit or blended finance8%3%12%
Harmonised definitions and certification4%1%10%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=22Capital grants for first plants (Borda 38%)
  • Subtask Bn=15Border carbon adjustment (CBAM) (Borda 38%)
  • Subtask Cn=16Lead markets or quotas for near-zero steel (Borda 42%)
  • Linkage and exchangen=12Lead markets or quotas for near-zero steel (Borda 42%)
  • Industry and policy referencen=20Lead markets or quotas for near-zero steel (Borda 38%)
  • Not yet assignedn=7Hydrogen production tax credits or subsidies (Borda 38%)
Q13 Low, best and high estimate n = 71

What share of global hydrogen-based DRI/HBI output will be traded internationally as merchant iron in 2040, rather than consumed at the site of production?

Give the lowest plausible and highest plausible values first, then your best estimate.

Median low

12.0

Median best

30.0

IQR 20.0–50.0

Median high

50.0

020406080100Interquartile range of best estimates: 20.0–50.0Median best estimate 30.0Expert: low 0, best 2, high 5Expert: low 0, best 2, high 5Expert: low 1, best 5, high 10Expert: low 0, best 8, high 15Expert: low 1, best 8, high 25Expert: low 5, best 8, high 10Expert: low 0, best 10, high 40Expert: low 5, best 10, high 20Expert: low 7, best 10, high 15Expert: low 5, best 11, high 20Expert: low 0, best 13, high 14Expert: low 5, best 15, high 25Expert: low 5, best 15, high 50Expert: low 10, best 15, high 20Expert: low 10, best 20, high 30Expert: low 10, best 20, high 30Expert: low 10, best 20, high 30Expert: low 10, best 20, high 30Expert: low 10, best 20, high 40Expert: low 10, best 20, high 30Expert: low 0, best 25, high 50Expert: low 2, best 25, high 50Expert: low 5, best 25, high 60Expert: low 8, best 25, high 51Expert: low 10, best 25, high 40Expert: low 10, best 25, high 50Expert: low 10, best 25, high 40Expert: low 10, best 25, high 40Expert: low 10, best 25, high 40Expert: low 12, best 25, high 40Expert: low 15, best 25, high 40Expert: low 20, best 25, high 40Expert: low 20, best 25, high 30Expert: low 20, best 26, high 26Expert: low 25, best 28, high 40Expert: low 3, best 30, high 36Expert: low 10, best 30, high 50Expert: low 20, best 30, high 50Expert: low 20, best 30, high 50Expert: low 25, best 30, high 30Expert: low 10, best 35, high 60Expert: low 15, best 35, high 75Expert: low 30, best 35, high 55Expert: low 24, best 38, high 54Expert: low 5, best 40, high 70Expert: low 10, best 40, high 75Expert: low 20, best 40, high 60Expert: low 25, best 40, high 60Expert: low 30, best 40, high 60Expert: low 30, best 45, high 50Expert: low 5, best 50, high 75Expert: low 20, best 50, high 74Expert: low 30, best 50, high 70Expert: low 30, best 50, high 60Expert: low 30, best 50, high 60Expert: low 30, best 50, high 70Expert: low 30, best 50, high 60Expert: low 45, best 50, high 55Expert: low 30, best 60, high 70Expert: low 50, best 60, high 80Expert: low 20, best 70, high 95Expert: low 20, best 70, high 95Expert: low 25, best 70, high 75Expert: low 25, best 70, high 90Expert: low 30, best 70, high 90Expert: low 50, best 70, high 80Expert: low 50, best 75, high 100Expert: low 65, best 75, high 85Expert: low 30, best 80, high 100Expert: low 80, best 90, high 100Expert: low 75, best 92, high 95

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: % of hydrogen-based DRI/HBI output

Unit: % of hydrogen-based DRI/HBI output
EstimatenMedianIQRRangeMean
Lowest plausible71126 to 280 to 8018.7
Best estimate713020 to 502 to 9235.5
Highest plausible715030 to 705 to 10050.9

By Task 52 affiliation (groups of at least 5)

  • Subtask An=20best 30% of hydrogen-based DRI/HBI output
  • Subtask Bn=13best 25% of hydrogen-based DRI/HBI output
  • Subtask Cn=12best 34% of hydrogen-based DRI/HBI output
  • Linkage and exchangen=8best 25% of hydrogen-based DRI/HBI output
  • Industry and policy referencen=13best 40% of hydrogen-based DRI/HBI output
  • Not yet assignedn=5best 25% of hydrogen-based DRI/HBI output
Q14 Single choice n = 96

On what basis should Task 52 define near-zero emission iron and steel?

An emissions-intensity threshold, technology-neutral: 52% (50) An emissions-intensity threshold, technology-… 52% (50) Eligible production routes: 4% (4) Eligible production routes 4% (4) A threshold combined with route eligibility rules: 31% (30) A threshold combined with route eligibility r… 31% (30) No single definition; report multiple metrics: 13% (12) No single definition; report multiple metrics 13% (12)
n = 96
OptionExpertsShare
An emissions-intensity threshold, technology-neutral5052%
Eligible production routes44%
A threshold combined with route eligibility rules3031%
No single definition; report multiple metrics1213%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=24An emissions-intensity threshold, technology-neutral 54%
  • Subtask Bn=16An emissions-intensity threshold, technology-neutral 56%
  • Subtask Cn=17An emissions-intensity threshold, technology-neutral 53%
  • Linkage and exchangen=12A threshold combined with route eligibility rules 58%
  • Industry and policy referencen=20An emissions-intensity threshold, technology-neutral 60%
  • Not yet assignedn=7An emissions-intensity threshold, technology-neutral 57%
Q15 Single choice n = 94

Which emissions scope boundary should that definition use?

Scope 1 and 2 only: 26% (24) Scope 1 and 2 only 26% (24) Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport): 62% (58) Scope 1 and 2 plus upstream Scope 3 (mining, … 62% (58) Full life cycle including downstream use: 11% (10) Full life cycle including downstream use 11% (10) Don't know: 2% (2) Don't know 2% (2)
n = 94
OptionExpertsShare
Scope 1 and 2 only2426%
Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport)5862%
Full life cycle including downstream use1011%
Don't know22%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=23Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport) 74%
  • Subtask Bn=16Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport) 50%
  • Subtask Cn=17Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport) 71%
  • Linkage and exchangen=12Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport) 67%
  • Industry and policy referencen=19Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport) 42%
  • Not yet assignedn=7Scope 1 and 2 plus upstream Scope 3 (mining, ore processing, transport) 71%
Q16 Single choice n = 92

For commercial-scale DRI projects reaching final investment decision in 2026 to 2028, which initial reductant configuration do you expect to be most common?

Hydrogen from commissioning: 2% (2) Hydrogen from commissioning 2% (2) Natural gas at start with a planned transition to hydrogen: 64% (59) Natural gas at start with a planned transitio… 64% (59) Mixed, differing by region: 33% (30) Mixed, differing by region 33% (30) Too early to say: 1% (1) Too early to say 1% (1)
n = 92
OptionExpertsShare
Hydrogen from commissioning22%
Natural gas at start with a planned transition to hydrogen5964%
Mixed, differing by region3033%
Too early to say11%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=24Natural gas at start with a planned transition to hydrogen 67%
  • Subtask Bn=15Natural gas at start with a planned transition to hydrogen 47%
  • Subtask Cn=17Natural gas at start with a planned transition to hydrogen 71%
  • Linkage and exchangen=12Mixed, differing by region 50%
  • Industry and policy referencen=18Natural gas at start with a planned transition to hydrogen 78%
  • Not yet assignedn=6Natural gas at start with a planned transition to hydrogen 83%
Technology readiness and process integration 47 experts · 13 questions
Q17 Choice grid n = 41

What is the current technology readiness of each route?

Use the IEA scale: TRL 1 to 4 concept and lab, 5 to 6 pilot, 7 to 8 demonstration and first-of-a-kind, 9 and above commercial with multiple plants.

Shaft-furnace DRI on 100% hydrogen, DR-… Shaft-furnace DRI on 100% hydrogen, DR-grade pellets · TRL 5 to 6: 20% (n=41)Shaft-furnace DRI on 100% hydrogen, DR-grade pellets · TRL 7 to 8: 71% (n=41)Shaft-furnace DRI on 100% hydrogen, DR-grade pellets · TRL 9 or above: 10% (n=41) Mean score (n=41)7.3 Shaft-furnace DRI on natural gas with 3… Shaft-furnace DRI on natural gas with 30% or more hydrogen blending · TRL 5 to 6: 2% (n=41)Shaft-furnace DRI on natural gas with 30% or more hydrogen blending · TRL 7 to 8: 39% (n=41)Shaft-furnace DRI on natural gas with 30% or more hydrogen blending · TRL 9 or above: 59% (n=41) Mean score (n=41)8.6 Fluidised-bed hydrogen reduction of fin… Fluidised-bed hydrogen reduction of fines · TRL 1 to 4: 13% (n=39)Fluidised-bed hydrogen reduction of fines · TRL 5 to 6: 51% (n=39)Fluidised-bed hydrogen reduction of fines · TRL 7 to 8: 36% (n=39) Mean score (n=39)5.9 Hydrogen DRI plus electric smelting fur… Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 1 to 4: 10% (n=40)Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 5 to 6: 48% (n=40)Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 7 to 8: 38% (n=40)Hydrogen DRI plus electric smelting furnace for lower-grade ores · TRL 9 or above: 5% (n=40) Mean score (n=40)6.2 Hydrogen plasma or smelting reduction Hydrogen plasma or smelting reduction · TRL 1 to 4: 45% (n=38)Hydrogen plasma or smelting reduction · TRL 5 to 6: 55% (n=38) Mean score (n=38)4.4 Molten oxide electrolysis Molten oxide electrolysis · TRL 1 to 4: 47% (n=38)Molten oxide electrolysis · TRL 5 to 6: 42% (n=38)Molten oxide electrolysis · TRL 7 to 8: 11% (n=38) Mean score (n=38)4.5 Low-temperature aqueous iron electrowin… Low-temperature aqueous iron electrowinning · TRL 1 to 4: 47% (n=38)Low-temperature aqueous iron electrowinning · TRL 5 to 6: 45% (n=38)Low-temperature aqueous iron electrowinning · TRL 7 to 8: 8% (n=38) Mean score (n=38)4.5 Blast furnace with hydrogen injection Blast furnace with hydrogen injection · TRL 1 to 4: 5% (n=41)Blast furnace with hydrogen injection · TRL 5 to 6: 20% (n=41)Blast furnace with hydrogen injection · TRL 7 to 8: 44% (n=41)Blast furnace with hydrogen injection · TRL 9 or above: 32% (n=41) Mean score (n=41)7.5 BF-BOF with CCUS BF-BOF with CCUS · TRL 1 to 4: 7% (n=41)BF-BOF with CCUS · TRL 5 to 6: 29% (n=41)BF-BOF with CCUS · TRL 7 to 8: 54% (n=41)BF-BOF with CCUS · TRL 9 or above: 10% (n=41) Mean score (n=41)6.8 Biomass or biochar-based ironmaking Biomass or biochar-based ironmaking · TRL 1 to 4: 7% (n=41)Biomass or biochar-based ironmaking · TRL 5 to 6: 15% (n=41)Biomass or biochar-based ironmaking · TRL 7 to 8: 39% (n=41)Biomass or biochar-based ironmaking · TRL 9 or above: 39% (n=41) Mean score (n=41)7.7
TRL 1 to 4TRL 5 to 6TRL 7 to 8TRL 9 or aboveNumber at right: mean score
Share of the experts answering each row
RownTRL 1 to 4TRL 5 to 6TRL 7 to 8TRL 9 or above
Shaft-furnace DRI on 100% hydrogen, DR-grade pellets410%20%71%10%
Shaft-furnace DRI on natural gas with 30% or more hydrogen blending410%2%39%59%
Fluidised-bed hydrogen reduction of fines3913%51%36%0%
Hydrogen DRI plus electric smelting furnace for lower-grade ores4010%48%38%5%
Hydrogen plasma or smelting reduction3845%55%0%0%
Molten oxide electrolysis3847%42%11%0%
Low-temperature aqueous iron electrowinning3847%45%8%0%
Blast furnace with hydrogen injection415%20%44%32%
BF-BOF with CCUS417%29%54%10%
Biomass or biochar-based ironmaking417%15%39%39%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=22Shaft-furnace DRI on 100% hydrogen, DR-grade pellets: TRL 7 to 8 73%
  • Industry and policy referencen=11Shaft-furnace DRI on 100% hydrogen, DR-grade pellets: TRL 7 to 8 55%
Q18 Choice grid n = 39

When will the first commercial-scale plant (at least 1 Mt/yr) be operating for each route?

Shaft-furnace DRI on 100% hydrogen Shaft-furnace DRI on 100% hydrogen · Already operating: 8% (n=39)Shaft-furnace DRI on 100% hydrogen · By 2028: 41% (n=39)Shaft-furnace DRI on 100% hydrogen · 2029 to 2032: 41% (n=39)Shaft-furnace DRI on 100% hydrogen · 2033 to 2036: 3% (n=39)Shaft-furnace DRI on 100% hydrogen · 2037 to 2040: 5% (n=39)Shaft-furnace DRI on 100% hydrogen · After 2040: 3% (n=39) Mean score (n=39)2029.5 Fluidised-bed hydrogen reduction of fin… Fluidised-bed hydrogen reduction of fines · Already operating: 6% (n=36)Fluidised-bed hydrogen reduction of fines · 2029 to 2032: 22% (n=36)Fluidised-bed hydrogen reduction of fines · 2033 to 2036: 56% (n=36)Fluidised-bed hydrogen reduction of fines · 2037 to 2040: 8% (n=36)Fluidised-bed hydrogen reduction of fines · After 2040: 8% (n=36) Mean score (n=36)2034.3 Hydrogen DRI plus electric smelting fur… Hydrogen DRI plus electric smelting furnace · Already operating: 3% (n=37)Hydrogen DRI plus electric smelting furnace · By 2028: 8% (n=37)Hydrogen DRI plus electric smelting furnace · 2029 to 2032: 43% (n=37)Hydrogen DRI plus electric smelting furnace · 2033 to 2036: 27% (n=37)Hydrogen DRI plus electric smelting furnace · 2037 to 2040: 16% (n=37)Hydrogen DRI plus electric smelting furnace · After 2040: 3% (n=37) Mean score (n=37)2032.8 Hydrogen plasma or smelting reduction Hydrogen plasma or smelting reduction · 2029 to 2032: 3% (n=35)Hydrogen plasma or smelting reduction · 2033 to 2036: 20% (n=35)Hydrogen plasma or smelting reduction · 2037 to 2040: 34% (n=35)Hydrogen plasma or smelting reduction · After 2040: 40% (n=35)Hydrogen plasma or smelting reduction · Never: 3% (n=35) Mean score (n=35)2040.1 Molten oxide electrolysis Molten oxide electrolysis · Already operating: 3% (n=35)Molten oxide electrolysis · By 2028: 3% (n=35)Molten oxide electrolysis · 2033 to 2036: 14% (n=35)Molten oxide electrolysis · 2037 to 2040: 20% (n=35)Molten oxide electrolysis · After 2040: 46% (n=35)Molten oxide electrolysis · Never: 14% (n=35) Mean score (n=35)2040.5 Low-temperature iron electrowinning Low-temperature iron electrowinning · 2029 to 2032: 3% (n=35)Low-temperature iron electrowinning · 2033 to 2036: 26% (n=35)Low-temperature iron electrowinning · 2037 to 2040: 17% (n=35)Low-temperature iron electrowinning · After 2040: 49% (n=35)Low-temperature iron electrowinning · Never: 6% (n=35) Mean score (n=35)2040.5
Already operatingBy 20282029 to 20322033 to 20362037 to 2040After 2040NeverNumber at right: mean score
Share of the experts answering each row
RownAlready operatingBy 20282029 to 20322033 to 20362037 to 2040After 2040Never
Shaft-furnace DRI on 100% hydrogen398%41%41%3%5%3%0%
Fluidised-bed hydrogen reduction of fines366%0%22%56%8%8%0%
Hydrogen DRI plus electric smelting furnace373%8%43%27%16%3%0%
Hydrogen plasma or smelting reduction350%0%3%20%34%40%3%
Molten oxide electrolysis353%3%0%14%20%46%14%
Low-temperature iron electrowinning350%0%3%26%17%49%6%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=20Shaft-furnace DRI on 100% hydrogen: 2029 to 2032 55%
  • Industry and policy referencen=10Shaft-furnace DRI on 100% hydrogen: By 2028 50%
Q19 Single choice n = 39

Which route will process most low-grade hematite ore into hydrogen-based iron by 2035?

Beneficiation to DR-grade pellets, then shaft-furnace DRI-EAF: 15% (6) Beneficiation to DR-grade pellets, then shaft… 15% (6) Shaft-furnace DRI on lower-grade pellets, then electric smelting furnace and BOF: 41% (16) Shaft-furnace DRI on lower-grade pellets, the… 41% (16) Fluidised-bed reduction of fines, then electric smelting furnace: 10% (4) Fluidised-bed reduction of fines, then electr… 10% (4) Direct smelting reduction or hydrogen plasma on fines: 8% (3) Direct smelting reduction or hydrogen plasma … 8% (3) Low-grade ore stays in BF-BOF, abated with CCUS: 3% (1) Low-grade ore stays in BF-BOF, abated with CC… 3% (1) Low-grade ore will largely not be used for hydrogen-based iron by 2035: 23% (9) Low-grade ore will largely not be used for hy… 23% (9)

Self-rated confidence: mean 3.5 of 5

n = 39
OptionExpertsShare
Beneficiation to DR-grade pellets, then shaft-furnace DRI-EAF615%
Shaft-furnace DRI on lower-grade pellets, then electric smelting furnace and BOF1641%
Fluidised-bed reduction of fines, then electric smelting furnace410%
Direct smelting reduction or hydrogen plasma on fines38%
Low-grade ore stays in BF-BOF, abated with CCUS13%
Low-grade ore will largely not be used for hydrogen-based iron by 2035923%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=21Shaft-furnace DRI on lower-grade pellets, then electric smelting furnace and BOF 48%
  • Subtask Bn=5Shaft-furnace DRI on lower-grade pellets, then electric smelting furnace and BOF 40%
  • Industry and policy referencen=8Shaft-furnace DRI on lower-grade pellets, then electric smelting furnace and BOF 38%
Q20 Low, best and high estimate n = 33

What total electricity input per tonne of crude steel will a best-in-class hydrogen-based DRI-EAF plant need in 2035, including electrolysis but excluding mining and pelletising?

Published estimates today span roughly 3.5 to 5 MWh/t. Give lowest plausible, highest plausible and best estimate.

Median low

3.3

Median best

4.0

IQR 3.5–4.5

Median high

5.0

12.43.85.26.68Interquartile range of best estimates: 3.5–4.5Median best estimate 4.0Expert: low 1, best 2, high 3Expert: low 1, best 3, high 5Expert: low 3, best 3.1, high 4.9Expert: low 3.3, best 3.4, high 4Expert: low 2.5, best 3.5, high 4.5Expert: low 3, best 3.5, high 5Expert: low 3, best 3.5, high 4Expert: low 3, best 3.5, high 4Expert: low 3, best 3.5, high 5Expert: low 3, best 3.5, high 4.5Expert: low 3, best 3.7, high 4.5Expert: low 3.5, best 3.8, high 4.5Expert: low 3.5, best 3.9, high 4.5Expert: low 3, best 4, high 5Expert: low 3, best 4, high 4.5Expert: low 3.5, best 4, high 5.5Expert: low 3.5, best 4, high 4.5Expert: low 3.5, best 4, high 5Expert: low 3.3, best 4.2, high 5.5Expert: low 3, best 4.4, high 5.5Expert: low 4.3, best 4.4, high 4.6Expert: low 1, best 4.5, high 5.8Expert: low 3.5, best 4.5, high 5Expert: low 3.5, best 4.5, high 5Expert: low 3.5, best 4.5, high 5.5Expert: low 3.7, best 4.5, high 6.9Expert: low 4, best 4.5, high 5.5Expert: low 4, best 4.5, high 5.5Expert: low 3, best 4.6, high 5.5Expert: low 3, best 5, high 7Expert: low 4, best 5, high 6Expert: low 3.3, best 5.5, high 7.8Expert: low 4.5, best 6, high 6.4

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: MWh per tonne of crude steel

Unit: MWh per tonne of crude steel
EstimatenMedianIQRRangeMean
Lowest plausible333.33.0 to 3.51.0 to 4.53.15
Best estimate334.03.5 to 4.52.0 to 6.04.08
Highest plausible335.04.5 to 5.53.0 to 7.85.13

By Task 52 affiliation (groups of at least 5)

  • Subtask An=16best 4.3 MWh per tonne of crude steel
  • Subtask Bn=5best 3.7 MWh per tonne of crude steel
  • Industry and policy referencen=7best 3.5 MWh per tonne of crude steel
Q21 Ranked top 3 n = 41

Rank the three most important technical barriers to scaling hydrogen-based DRI from demonstration to multiple commercial plants.

Hydrogen supply variability and buffering: 73% · 1st 63% Hydrogen supply variability and buffering 73% · 1st 63% Availability and consistency of DR-grade pellets: 29% · 1st 10% Availability and consistency of DR-grade pell… 29% · 1st 10% Heat balance and gas preheating with hydrogen: 26% · 1st 15% Heat balance and gas preheating with hydrogen 26% · 1st 15% Pellet sticking, fines generation and shaft-furnace behaviour with hydrogen: 15% · 1st 0% Pellet sticking, fines generation and shaft-f… 15% · 1st 0% DRI carbon content and metallisation for EAF steelmaking: 15% · 1st 2% DRI carbon content and metallisation for EAF … 15% · 1st 2% Hydrogen safety and materials compatibility: 15% · 1st 2% Hydrogen safety and materials compatibility 15% · 1st 2% Process control and operating experience: 12% · 1st 5% Process control and operating experience 12% · 1st 5% Scale-up of fluidised-bed reactors: 11% · 1st 2% Scale-up of fluidised-bed reactors 11% · 1st 2% Equipment supply chain and EPC capacity: 3% · 1st 0% Equipment supply chain and EPC capacity 3% · 1st 0%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 41 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Hydrogen supply variability and buffering73%63%83%
Availability and consistency of DR-grade pellets29%10%44%
Heat balance and gas preheating with hydrogen26%15%37%
Pellet sticking, fines generation and shaft-furnace behaviour with hydrogen15%0%29%
DRI carbon content and metallisation for EAF steelmaking15%2%32%
Hydrogen safety and materials compatibility15%2%27%
Process control and operating experience12%5%20%
Scale-up of fluidised-bed reactors11%2%20%
Equipment supply chain and EPC capacity3%0%10%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=21Hydrogen supply variability and buffering (Borda 79%)
  • Subtask Bn=5Hydrogen supply variability and buffering (Borda 100%)
  • Industry and policy referencen=9Hydrogen supply variability and buffering (Borda 67%)
Q22 Rating, 1 (Minor) to 5 (Critical) n = 36

How serious is each challenge for EAF-based steelmaking on hydrogen DRI?

Gangue and slag volume from lower-grade… Gangue and slag volume from lower-grade DRI · 2: 9% (n=35)Gangue and slag volume from lower-grade DRI · 3: 11% (n=35)Gangue and slag volume from lower-grade DRI · 4: 26% (n=35)Gangue and slag volume from lower-grade DRI · 5: 54% (n=35) Mean score (n=35)4.3 Scrap residuals (copper, tin) limiting … Scrap residuals (copper, tin) limiting product mix · 1: 3% (n=35)Scrap residuals (copper, tin) limiting product mix · 2: 20% (n=35)Scrap residuals (copper, tin) limiting product mix · 3: 29% (n=35)Scrap residuals (copper, tin) limiting product mix · 4: 20% (n=35)Scrap residuals (copper, tin) limiting product mix · 5: 29% (n=35) Mean score (n=35)3.5 EAF productivity, electrode and power d… EAF productivity, electrode and power demand · 1: 6% (n=33)EAF productivity, electrode and power demand · 2: 15% (n=33)EAF productivity, electrode and power demand · 3: 39% (n=33)EAF productivity, electrode and power demand · 4: 30% (n=33)EAF productivity, electrode and power demand · 5: 9% (n=33) Mean score (n=33)3.2 Quality for high-grade flat products Quality for high-grade flat products · 1: 3% (n=35)Quality for high-grade flat products · 2: 31% (n=35)Quality for high-grade flat products · 3: 23% (n=35)Quality for high-grade flat products · 4: 29% (n=35)Quality for high-grade flat products · 5: 14% (n=35) Mean score (n=35)3.2 Integration of an electric smelting fur… Integration of an electric smelting furnace before the BOF · 1: 20% (n=35)Integration of an electric smelting furnace before the BOF · 2: 9% (n=35)Integration of an electric smelting furnace before the BOF · 3: 34% (n=35)Integration of an electric smelting furnace before the BOF · 4: 34% (n=35)Integration of an electric smelting furnace before the BOF · 5: 3% (n=35) Mean score (n=35)2.9 HBI reoxidation, handling and transport… HBI reoxidation, handling and transport safety · 1: 11% (n=35)HBI reoxidation, handling and transport safety · 2: 37% (n=35)HBI reoxidation, handling and transport safety · 3: 43% (n=35)HBI reoxidation, handling and transport safety · 4: 6% (n=35)HBI reoxidation, handling and transport safety · 5: 3% (n=35) Mean score (n=35)2.5 Operating skills and workforce transiti… Operating skills and workforce transition · 1: 18% (n=33)Operating skills and workforce transition · 2: 36% (n=33)Operating skills and workforce transition · 3: 30% (n=33)Operating skills and workforce transition · 4: 12% (n=33)Operating skills and workforce transition · 5: 3% (n=33) Mean score (n=33)2.5
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Gangue and slag volume from lower-grade DRI354.265.04.0 to 5.0
Scrap residuals (copper, tin) limiting product mix353.513.03.0 to 5.0
EAF productivity, electrode and power demand333.213.03.0 to 4.0
Quality for high-grade flat products353.203.02.0 to 4.0
Integration of an electric smelting furnace before the BOF352.913.02.0 to 4.0
HBI reoxidation, handling and transport safety352.513.02.0 to 3.0
Operating skills and workforce transition332.452.02.0 to 3.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=20Gangue and slag volume from lower-grade DRI 4.5
  • Industry and policy referencen=9Gangue and slag volume from lower-grade DRI 3.7
Q23 Multiple choice, up to 3 n = 39

Which data are hardest to obtain for assessing hydrogen-based ironmaking technologies? Select up to three.

Operating costs and availability of demonstration plants: 77% Operating costs and availability of demonstra… 77% Realised capex of demonstration and first commercial plants: 69% Realised capex of demonstration and first com… 69% Measured energy and hydrogen consumption: 38% Measured energy and hydrogen consumption 38% Reliability, maintenance and failure data: 33% Reliability, maintenance and failure data 33% Ore and pellet performance in hydrogen reduction: 21% Ore and pellet performance in hydrogen reduct… 21% DRI/HBI product quality and downstream performance: 21% DRI/HBI product quality and downstream perfor… 21% Verified emissions accounting: 15% Verified emissions accounting 15% Project pipeline status and cancellations: 13% Project pipeline status and cancellations 13% Other (please specify): 3% Other (please specify) 3%

Share of respondents selecting each option.

n = 39 · each expert could select up to 3
OptionExpertsShare selecting
Operating costs and availability of demonstration plants3077%
Realised capex of demonstration and first commercial plants2769%
Measured energy and hydrogen consumption1538%
Reliability, maintenance and failure data1333%
Ore and pellet performance in hydrogen reduction821%
DRI/HBI product quality and downstream performance821%
Verified emissions accounting615%
Project pipeline status and cancellations513%
Other (please specify)13%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=19Operating costs and availability of demonstration plants 79%
  • Industry and policy referencen=10Realised capex of demonstration and first commercial plants 80%
Q24 Rating, 1 (Not a real problem) to 5 (Major unresolved risk) n = 32

Based on publicly available evidence, how important is each technical challenge for scaling hydrogen-based DRI?

Hydrogen supply reliability, pressure a… Hydrogen supply reliability, pressure and purity at the plant boundary · 1: 3% (n=31)Hydrogen supply reliability, pressure and purity at the plant boundary · 2: 10% (n=31)Hydrogen supply reliability, pressure and purity at the plant boundary · 3: 13% (n=31)Hydrogen supply reliability, pressure and purity at the plant boundary · 4: 45% (n=31)Hydrogen supply reliability, pressure and purity at the plant boundary · 5: 29% (n=31) Mean score (n=31)3.9 Pellet sticking, clustering or fines ge… Pellet sticking, clustering or fines generation in the shaft · 1: 6% (n=32)Pellet sticking, clustering or fines generation in the shaft · 2: 9% (n=32)Pellet sticking, clustering or fines generation in the shaft · 3: 28% (n=32)Pellet sticking, clustering or fines generation in the shaft · 4: 38% (n=32)Pellet sticking, clustering or fines generation in the shaft · 5: 19% (n=32) Mean score (n=32)3.5 Electric smelting furnace slag chemistr… Electric smelting furnace slag chemistry and arc stability at scale · 1: 3% (n=30)Electric smelting furnace slag chemistry and arc stability at scale · 2: 13% (n=30)Electric smelting furnace slag chemistry and arc stability at scale · 3: 30% (n=30)Electric smelting furnace slag chemistry and arc stability at scale · 4: 37% (n=30)Electric smelting furnace slag chemistry and arc stability at scale · 5: 17% (n=30) Mean score (n=30)3.5 Behaviour of magnetite or lower-grade o… Behaviour of magnetite or lower-grade ore in fluidised or flash reduction · 1: 4% (n=28)Behaviour of magnetite or lower-grade ore in fluidised or flash reduction · 2: 18% (n=28)Behaviour of magnetite or lower-grade ore in fluidised or flash reduction · 3: 25% (n=28)Behaviour of magnetite or lower-grade ore in fluidised or flash reduction · 4: 36% (n=28)Behaviour of magnetite or lower-grade ore in fluidised or flash reduction · 5: 18% (n=28) Mean score (n=28)3.5 Meeting metallisation and carbon-conten… Meeting metallisation and carbon-content targets at the same time · 1: 3% (n=32)Meeting metallisation and carbon-content targets at the same time · 2: 16% (n=32)Meeting metallisation and carbon-content targets at the same time · 3: 28% (n=32)Meeting metallisation and carbon-content targets at the same time · 4: 41% (n=32)Meeting metallisation and carbon-content targets at the same time · 5: 13% (n=32) Mean score (n=32)3.4 Refractory and burner-material wear und… Refractory and burner-material wear under high-hydrogen operation · 1: 7% (n=28)Refractory and burner-material wear under high-hydrogen operation · 2: 14% (n=28)Refractory and burner-material wear under high-hydrogen operation · 3: 43% (n=28)Refractory and burner-material wear under high-hydrogen operation · 4: 25% (n=28)Refractory and burner-material wear under high-hydrogen operation · 5: 11% (n=28) Mean score (n=28)3.2 Availability shortfall against design i… Availability shortfall against design in the first two years · 1: 13% (n=24)Availability shortfall against design in the first two years · 2: 13% (n=24)Availability shortfall against design in the first two years · 3: 33% (n=24)Availability shortfall against design in the first two years · 4: 33% (n=24)Availability shortfall against design in the first two years · 5: 8% (n=24) Mean score (n=24)3.1 DRI or HBI reoxidation during handling,… DRI or HBI reoxidation during handling, storage and transport · 1: 6% (n=31)DRI or HBI reoxidation during handling, storage and transport · 2: 32% (n=31)DRI or HBI reoxidation during handling, storage and transport · 3: 39% (n=31)DRI or HBI reoxidation during handling, storage and transport · 4: 23% (n=31) Mean score (n=31)2.8
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Hydrogen supply reliability, pressure and purity at the plant boundary313.874.03.5 to 5.0
Pellet sticking, clustering or fines generation in the shaft323.534.03.0 to 4.0
Electric smelting furnace slag chemistry and arc stability at scale303.504.03.0 to 4.0
Behaviour of magnetite or lower-grade ore in fluidised or flash reduction283.464.03.0 to 4.0
Meeting metallisation and carbon-content targets at the same time323.444.03.0 to 4.0
Refractory and burner-material wear under high-hydrogen operation283.183.03.0 to 4.0
Availability shortfall against design in the first two years243.133.02.8 to 4.0
DRI or HBI reoxidation during handling, storage and transport312.773.02.0 to 3.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=19Hydrogen supply reliability, pressure and purity at the plant boundary 3.9
  • Industry and policy referencen=7Hydrogen supply reliability, pressure and purity at the plant boundary 4.1
Q25 Low, best and high estimate n = 33

For a representative first-of-a-kind commercial-scale hydrogen shaft-furnace DRI plant, what capacity factor is it reasonable to expect over the first 24 months of commercial operation?

Capacity factor means annual output divided by nameplate annual capacity. Judge from public evidence and general experience of first-of-a-kind process plants; do not use non-public information about any specific plant.

Median low

40.0

Median best

65.0

IQR 55.0–70.0

Median high

80.0

020406080100Interquartile range of best estimates: 55.0–70.0Median best estimate 65.0Expert: low 5, best 10, high 15Expert: low 5, best 30, high 90Expert: low 15, best 50, high 80Expert: low 20, best 50, high 70Expert: low 25, best 50, high 75Expert: low 30, best 50, high 80Expert: low 35, best 50, high 80Expert: low 40, best 50, high 60Expert: low 30, best 55, high 70Expert: low 40, best 55, high 70Expert: low 40, best 60, high 85Expert: low 40, best 60, high 80Expert: low 40, best 60, high 70Expert: low 40, best 60, high 80Expert: low 50, best 60, high 100Expert: low 50, best 60, high 80Expert: low 30, best 65, high 100Expert: low 40, best 65, high 80Expert: low 40, best 65, high 85Expert: low 50, best 65, high 80Expert: low 30, best 70, high 80Expert: low 50, best 70, high 80Expert: low 55, best 70, high 85Expert: low 60, best 70, high 80Expert: low 60, best 70, high 80Expert: low 60, best 70, high 80Expert: low 65, best 75, high 92Expert: low 50, best 80, high 100Expert: low 60, best 80, high 90Expert: low 60, best 85, high 90Expert: low 66, best 85, high 100Expert: low 50, best 90, high 100Expert: low 70, best 90, high 100

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: % of nameplate capacity

Unit: % of nameplate capacity
EstimatenMedianIQRRangeMean
Lowest plausible334030 to 555 to 7042.5
Best estimate336555 to 7010 to 9062.9
Highest plausible338080 to 9015 to 10081.4

By Task 52 affiliation (groups of at least 5)

  • Subtask An=16best 63% of nameplate capacity
  • Industry and policy referencen=9best 65% of nameplate capacity
Q26 Single choice n = 43

By 2035, what will be the dominant source of the carbon added to hydrogen-based DRI to reach the carburisation needed for EAF steelmaking?

Natural gas or methane injection, used pragmatically despite the emissions: 65% (28) Natural gas or methane injection, used pragma… 65% (28) Biochar or other solid biogenic carbon: 16% (7) Biochar or other solid biogenic carbon 16% (7) Biogenic CO2 or biomethane: 5% (2) Biogenic CO2 or biomethane 5% (2) A mix with no single dominant source: 14% (6) A mix with no single dominant source 14% (6) Process design minimises the need for added carbon: 0% (0) Process design minimises the need for added c… 0% (0)
n = 43
OptionExpertsShare
Natural gas or methane injection, used pragmatically despite the emissions2865%
Biochar or other solid biogenic carbon716%
Biogenic CO2 or biomethane25%
A mix with no single dominant source614%
Process design minimises the need for added carbon00%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=21Natural gas or methane injection, used pragmatically despite the emissions 62%
  • Subtask Bn=5Natural gas or methane injection, used pragmatically despite the emissions 60%
  • Industry and policy referencen=12Natural gas or methane injection, used pragmatically despite the emissions 58%
Q27 Single choice n = 39

For flat, high-quality steel made from hydrogen DRI via an electric smelting furnace, which melting vessel will take most of the global tonnage by 2035?

BOF dominates; smelter hot metal is treated like blast-furnace hot metal: 41% (16) BOF dominates; smelter hot metal is treated l… 41% (16) EAF dominates; smelter hot metal is blended with scrap: 36% (14) EAF dominates; smelter hot metal is blended w… 36% (14) Roughly even, differing by region and product: 15% (6) Roughly even, differing by region and product 15% (6) Neither; smelter-route tonnage stays marginal by 2035: 8% (3) Neither; smelter-route tonnage stays marginal… 8% (3)
n = 39
OptionExpertsShare
BOF dominates; smelter hot metal is treated like blast-furnace hot metal1641%
EAF dominates; smelter hot metal is blended with scrap1436%
Roughly even, differing by region and product615%
Neither; smelter-route tonnage stays marginal by 203538%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=20BOF dominates; smelter hot metal is treated like blast-furnace hot metal 50%
  • Industry and policy referencen=10EAF dominates; smelter hot metal is blended with scrap 60%
Q28 Single choice n = 34

For hydrogen DRI plants commissioned between 2028 and 2035 that are co-located with their own melt shop, which product-handling approach will be more common?

Hot DRI charged directly to the melter, no HBI step: 71% (24) Hot DRI charged directly to the melter, no HB… 71% (24) Hot-transport-vessel charging with partial cooling: 18% (6) Hot-transport-vessel charging with partial co… 18% (6) Cooled to HBI even at co-located sites, for flexibility and storage: 9% (3) Cooled to HBI even at co-located sites, for f… 9% (3) Roughly even split: 3% (1) Roughly even split 3% (1)
n = 34
OptionExpertsShare
Hot DRI charged directly to the melter, no HBI step2471%
Hot-transport-vessel charging with partial cooling618%
Cooled to HBI even at co-located sites, for flexibility and storage39%
Roughly even split13%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=18Hot DRI charged directly to the melter, no HBI step 72%
  • Industry and policy referencen=8Hot DRI charged directly to the melter, no HBI step 63%
Q29 Low, best and high estimate n = 16

What will total water consumption be for a best-in-class hydrogen-based DRI-EAF plant in 2035, including electrolysis feed water and process cooling but excluding mining and beneficiation, per tonne of crude steel?

Electrolysis alone needs roughly 9 litres of water per kilogram of hydrogen.

Median low

5.0

Median best

8.7

IQR 3.9–13.0

Median high

12.8

048121620Interquartile range of best estimates: 3.9–13.0Median best estimate 8.7Expert: low 0.8, best 1.9, high 3Expert: low 1, best 2, high 4Expert: low 2, best 2.5, high 3Expert: low 1, best 3.1, high 4Expert: low 3.5, best 4.2, high 5Expert: low 5, best 6.5, high 10.5Expert: low 6, best 7, high 8Expert: low 0.3, best 7.3, high 10.5Expert: low 1, best 10, high 20Expert: low 5, best 10, high 15Expert: low 11, best 13, high 15Expert: low 11, best 13, high 15Expert: low 12.1, best 13, high 15Expert: low 12, best 14, high 17Expert: low 8, best 15, high 18Expert: low 12, best 15, high 16

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: m³ water per tonne of crude steel

Unit: m³ water per tonne of crude steel
EstimatenMedianIQRRangeMean
Lowest plausible165.01.0 to 11.00.3 to 12.15.73
Best estimate168.73.9 to 13.01.9 to 15.08.59
Highest plausible1612.84.8 to 15.33.0 to 20.011.19

By Task 52 affiliation (groups of at least 5)

  • Subtask An=10best 11.5 m³ water per tonne of crude steel
Process flexibility 34 experts · 8 questions
Q30 Single choice n = 20

What minimum stable production rate, as a share of nameplate, will a hydrogen shaft-furnace DRI plant sustain in 2030 without damaging refractories or product quality?

Consider sustained operation for days, not brief dips. Today's natural-gas shaft furnaces typically run close to nameplate.

Below 30% of nameplate: 10% (2) Below 30% of nameplate 10% (2) 30 to 50%: 5% (1) 30 to 50% 5% (1) 50 to 70%: 35% (7) 50 to 70% 35% (7) 70 to 90%: 35% (7) 70 to 90% 35% (7) No meaningful turndown; the plant must run near 100%: 15% (3) No meaningful turndown; the plant must run ne… 15% (3)

Mean score 68.0, median 70.0 · Self-rated confidence: mean 3.8 of 5

n = 20 · mean score 68.00, median 70.0
OptionExpertsShare
Below 30% of nameplate210%
30 to 50%15%
50 to 70%735%
70 to 90%735%
No meaningful turndown; the plant must run near 100%315%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=1370 to 90% 46%
  • Industry and policy referencen=550 to 70% 60%
Q31 Single choice n = 22

How quickly will a hydrogen-based DRI plant be able to move between 50% and 100% of nameplate production in 2030?

Under 1 hour: 9% (2) Under 1 hour 9% (2) 1 to 6 hours: 32% (7) 1 to 6 hours 32% (7) 6 to 24 hours: 36% (8) 6 to 24 hours 36% (8) More than 24 hours: 18% (4) More than 24 hours 18% (4) Only through planned weekly or seasonal scheduling: 5% (1) Only through planned weekly or seasonal sched… 5% (1)

Mean score 16.1, median 15.0

n = 22 · mean score 16.07, median 15.0
OptionExpertsShare
Under 1 hour29%
1 to 6 hours732%
6 to 24 hours836%
More than 24 hours418%
Only through planned weekly or seasonal scheduling15%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=146 to 24 hours 43%
  • Industry and policy referencen=51 to 6 hours 40%
Q32 Allocation of 100 points n = 20

Where should the flexibility in a renewable-powered hydrogen DRI system sit? Allocate 100 points across the options by the share of load-following each should provide in a least-cost 2035 design.

0%20%40%60%80%100% Electrolyser modulation: median 25.0%, IQR 13.5–34.5, range 9.0–40.0 (n=19) Electrolyser modulation 25 Hydrogen storage (pipes, tanks, caverns): median 25.0%, IQR 10.0–29.1, range 3.0–74.2 (n=18) Hydrogen storage (pipes, tanks, cav… 25 DRI plant load modulation: median 10.0%, IQR 5.3–20.0, range 3.5–25.0 (n=17) DRI plant load modulation 10 DRI/HBI inventory buffering before the EAF: median 7.5%, IQR 5.0–16.6, range 4.7–25.0 (n=16) DRI/HBI inventory buffering before … 8 EAF scheduling and melt-shop flexibility: median 10.0%, IQR 5.0–10.4, range 0.0–25.0 (n=14) EAF scheduling and melt-shop flexib… 10 Grid imports and firm power contracts: median 10.0%, IQR 7.0–20.0, range 0.0–40.0 (n=17) Grid imports and firm power contrac… 10 Batteries or other electricity storage: median 11.0%, IQR 10.0–25.0, range 0.0–100.0 (n=17) Batteries or other electricity stor… 11

Box: interquartile range · thick line: median · whiskers: min–max · number: median

Points out of 100 · reference values are the baseline shown to experts, where one was given
OptionnMedianIQRRangeReference
Electrolyser modulation1925.013.5 to 34.59 to 40–
Hydrogen storage (pipes, tanks, caverns)1825.010.0 to 29.13 to 74–
DRI plant load modulation1710.05.3 to 20.04 to 25–
DRI/HBI inventory buffering before the EAF167.55.0 to 16.65 to 25–
EAF scheduling and melt-shop flexibility1410.05.0 to 10.40 to 25–
Grid imports and firm power contracts1710.07.0 to 20.00 to 40–
Batteries or other electricity storage1711.010.0 to 25.00 to 100–

By Task 52 affiliation (groups of at least 5)

  • Subtask An=12Hydrogen storage (pipes, tanks, caverns) 25
  • Industry and policy referencen=6Electrolyser modulation 28
Q33 Single choice n = 24

What is the lowest iron content of pellet feed that a hydrogen shaft furnace will use economically by 2030, assuming an electric smelting furnace or adapted EAF downstream?

DR-grade pellets today are 67% Fe or better; blast-furnace-grade pellets are typically 62 to 65% Fe.

67% Fe or better only (DR-grade): 8% (2) 67% Fe or better only (DR-grade) 8% (2) 65 to 67% Fe: 38% (9) 65 to 67% Fe 38% (9) 62 to 65% Fe: 46% (11) 62 to 65% Fe 46% (11) Below 62% Fe: 8% (2) Below 62% Fe 8% (2)

Mean score 64.5, median 63.5 · Self-rated confidence: mean 3.8 of 5

n = 24 · mean score 64.48, median 63.5
OptionExpertsShare
67% Fe or better only (DR-grade)28%
65 to 67% Fe938%
62 to 65% Fe1146%
Below 62% Fe28%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=1662 to 65% Fe 63%
  • Industry and policy referencen=565 to 67% Fe 60%
Q34 Rating, 1 (Not feasible) to 5 (Routine) n = 21

How feasible is each feed-flexibility capability at commercial scale by 2035?

Varying the scrap-to-DRI ratio in the E… Varying the scrap-to-DRI ratio in the EAF with market prices · 4: 52% (n=21)Varying the scrap-to-DRI ratio in the EAF with market prices · 5: 48% (n=21) Mean score (n=21)4.5 Accepting higher gangue with an electri… Accepting higher gangue with an electric smelting furnace · 2: 10% (n=21)Accepting higher gangue with an electric smelting furnace · 3: 29% (n=21)Accepting higher gangue with an electric smelting furnace · 4: 38% (n=21)Accepting higher gangue with an electric smelting furnace · 5: 24% (n=21) Mean score (n=21)3.8 Switching pellet suppliers without proc… Switching pellet suppliers without process changes · 1: 5% (n=21)Switching pellet suppliers without process changes · 2: 19% (n=21)Switching pellet suppliers without process changes · 3: 19% (n=21)Switching pellet suppliers without process changes · 4: 29% (n=21)Switching pellet suppliers without process changes · 5: 29% (n=21) Mean score (n=21)3.6 Running on magnetite concentrates as we… Running on magnetite concentrates as well as hematite · 2: 10% (n=20)Running on magnetite concentrates as well as hematite · 3: 40% (n=20)Running on magnetite concentrates as well as hematite · 4: 35% (n=20)Running on magnetite concentrates as well as hematite · 5: 15% (n=20) Mean score (n=20)3.5 Blending DR-grade and BF-grade pellets … Blending DR-grade and BF-grade pellets on the fly · 1: 10% (n=21)Blending DR-grade and BF-grade pellets on the fly · 2: 10% (n=21)Blending DR-grade and BF-grade pellets on the fly · 3: 24% (n=21)Blending DR-grade and BF-grade pellets on the fly · 4: 33% (n=21)Blending DR-grade and BF-grade pellets on the fly · 5: 24% (n=21) Mean score (n=21)3.5 Using fines directly in fluidised-bed r… Using fines directly in fluidised-bed reduction · 1: 6% (n=18)Using fines directly in fluidised-bed reduction · 2: 17% (n=18)Using fines directly in fluidised-bed reduction · 3: 17% (n=18)Using fines directly in fluidised-bed reduction · 4: 50% (n=18)Using fines directly in fluidised-bed reduction · 5: 11% (n=18) Mean score (n=18)3.4 Charging lump ore alongside pellets in … Charging lump ore alongside pellets in a hydrogen shaft furnace · 2: 25% (n=20)Charging lump ore alongside pellets in a hydrogen shaft furnace · 3: 30% (n=20)Charging lump ore alongside pellets in a hydrogen shaft furnace · 4: 30% (n=20)Charging lump ore alongside pellets in a hydrogen shaft furnace · 5: 15% (n=20) Mean score (n=20)3.4
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Varying the scrap-to-DRI ratio in the EAF with market prices214.484.04.0 to 5.0
Accepting higher gangue with an electric smelting furnace213.764.03.0 to 4.0
Switching pellet suppliers without process changes213.574.03.0 to 5.0
Running on magnetite concentrates as well as hematite203.553.53.0 to 4.0
Blending DR-grade and BF-grade pellets on the fly213.524.03.0 to 4.0
Using fines directly in fluidised-bed reduction183.444.03.0 to 4.0
Charging lump ore alongside pellets in a hydrogen shaft furnace203.353.02.8 to 4.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=14Varying the scrap-to-DRI ratio in the EAF with market prices 4.4
Q35 Ranked top 3 n = 27

How will most hydrogen-ready DRI plants use their reductant flexibility in 2035? Rank the three most likely modes of operation, most likely first.

Run mostly on natural gas with hydrogen used only when subsidised: 56% · 1st 37% Run mostly on natural gas with hydrogen used … 56% · 1st 37% Vary the hydrogen share seasonally or by contract period: 53% · 1st 19% Vary the hydrogen share seasonally or by cont… 53% · 1st 19% Run a fixed hydrogen-natural gas blend set by certification thresholds: 37% · 1st 11% Run a fixed hydrogen-natural gas blend set by… 37% · 1st 11% Run on hydrogen only; natural gas kept for start-up and back-up: 31% · 1st 22% Run on hydrogen only; natural gas kept for st… 31% · 1st 22% Switch between hydrogen and natural gas from hour to hour with prices: 23% · 1st 11% Switch between hydrogen and natural gas from … 23% · 1st 11%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 27 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Run mostly on natural gas with hydrogen used only when subsidised56%37%67%
Vary the hydrogen share seasonally or by contract period53%19%89%
Run a fixed hydrogen-natural gas blend set by certification thresholds37%11%63%
Run on hydrogen only; natural gas kept for start-up and back-up31%22%44%
Switch between hydrogen and natural gas from hour to hour with prices23%11%37%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=17Vary the hydrogen share seasonally or by contract period (Borda 57%)
  • Industry and policy referencen=7Run mostly on natural gas with hydrogen used only when subsidised (Borda 67%)
Q36 Ranked top 3 n = 28

Rank the three forms of flexibility that will matter most for the competitiveness of hydrogen-based ironmaking.

Locational flexibility (moving reduction to where power is cheap): 49% · 1st 36% Locational flexibility (moving reduction to w… 49% · 1st 36% Ore grade and feed flexibility: 45% · 1st 21% Ore grade and feed flexibility 45% · 1st 21% Hydrogen and natural-gas reductant switching: 40% · 1st 21% Hydrogen and natural-gas reductant switching 40% · 1st 21% Electrical load flexibility (turndown and ramping): 32% · 1st 11% Electrical load flexibility (turndown and ram… 32% · 1st 11% Scrap-ratio flexibility in the EAF: 15% · 1st 7% Scrap-ratio flexibility in the EAF 15% · 1st 7% Output and inventory flexibility (stockpiling HBI): 10% · 1st 0% Output and inventory flexibility (stockpiling… 10% · 1st 0% Product flexibility (hot DRI, HBI, hot metal via ESF): 8% · 1st 4% Product flexibility (hot DRI, HBI, hot metal … 8% · 1st 4%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 28 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Locational flexibility (moving reduction to where power is cheap)49%36%57%
Ore grade and feed flexibility45%21%75%
Hydrogen and natural-gas reductant switching40%21%57%
Electrical load flexibility (turndown and ramping)32%11%50%
Scrap-ratio flexibility in the EAF15%7%21%
Output and inventory flexibility (stockpiling HBI)10%0%21%
Product flexibility (hot DRI, HBI, hot metal via ESF)8%4%18%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=16Ore grade and feed flexibility (Borda 58%)
  • Industry and policy referencen=8Hydrogen and natural-gas reductant switching (Borda 63%)
Q37 Single choice n = 24

For new hydrogen DRI plants powered substantially by variable renewables, which electrolyser-integration architecture will be more common by 2032?

Electrolysers directly coupled to renewable generation; the DRI plant absorbs the variability: 4% (1) Electrolysers directly coupled to renewable g… 4% (1) Battery buffering of electrolyser output to smooth hydrogen production: 21% (5) Battery buffering of electrolyser output to s… 21% (5) Hydrogen storage between electrolyser and DRI plant: 38% (9) Hydrogen storage between electrolyser and DRI… 38% (9) Grid-connected electrolysers running near-continuously, decoupled from on-site renewables: 38% (9) Grid-connected electrolysers running near-con… 38% (9)
n = 24
OptionExpertsShare
Electrolysers directly coupled to renewable generation; the DRI plant absorbs the variability14%
Battery buffering of electrolyser output to smooth hydrogen production521%
Hydrogen storage between electrolyser and DRI plant938%
Grid-connected electrolysers running near-continuously, decoupled from on-site renewables938%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=14Hydrogen storage between electrolyser and DRI plant 43%
  • Industry and policy referencen=7Hydrogen storage between electrolyser and DRI plant 43%
Resources, supply chains and infrastructure 37 experts · 16 questions
Q38 Allocation of 100 points n = 30

Allocate global hydrogen-based DRI/HBI production in 2040 across producing regions (percent, summing to 100).

0%20%40%60%80%100% Australia: median 10.0%, IQR 7.3–20.0, range 0.0–40.0 (n=28) Australia 10 Middle East and North Africa: median 15.0%, IQR 10.0–22.9, range 0.0–44.0 (n=27) Middle East and North Africa 15 Brazil and rest of Latin America: median 10.0%, IQR 5.0–11.1, range 0.0–25.7 (n=25) Brazil and rest of Latin America 10 Sub-Saharan Africa: median 5.0%, IQR 1.0–5.0, range 0.0–11.1 (n=17) Sub-Saharan Africa 5 China: median 20.5%, IQR 14.8–32.0, range 5.0–64.0 (n=28) China 21 India: median 6.0%, IQR 4.0–13.5, range 0.0–21.2 (n=23) India 6 Japan, Korea and South-East Asia: median 5.0%, IQR 2.5–10.2, range 0.0–20.0 (n=23) Japan, Korea and South-East Asia 5 Europe: median 11.6%, IQR 6.5–25.0, range 0.0–75.0 (n=28) Europe 12 North America: median 10.0%, IQR 5.1–10.0, range 0.0–25.0 (n=26) North America 10 Other regions: median 1.0%, IQR 0.0–5.0, range 0.0–11.1 (n=17) Other regions 1

Box: interquartile range · thick line: median · whiskers: min–max · number: median

Points out of 100 · reference values are the baseline shown to experts, where one was given
OptionnMedianIQRRangeReference
Australia2810.07.3 to 20.00 to 40–
Middle East and North Africa2715.010.0 to 22.90 to 44–
Brazil and rest of Latin America2510.05.0 to 11.10 to 26–
Sub-Saharan Africa175.01.0 to 5.00 to 11–
China2820.514.8 to 32.05 to 64–
India236.04.0 to 13.50 to 21–
Japan, Korea and South-East Asia235.02.5 to 10.20 to 20–
Europe2811.66.5 to 25.00 to 75–
North America2610.05.1 to 10.00 to 25–
Other regions171.00.0 to 5.00 to 11–

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9China 21
  • Subtask Bn=11Middle East and North Africa 17
  • Industry and policy referencen=5Middle East and North Africa 35
Q39 Single choice n = 27

What share of today's seaborne iron ore trade (by mass) can be economically upgraded to DR-grade feed (67% Fe or better, low gangue) by 2035?

Less than 10%: 33% (9) Less than 10% 33% (9) 10 to 25%: 44% (12) 10 to 25% 44% (12) 25 to 50%: 22% (6) 25 to 50% 22% (6) More than 50%: 0% (0) More than 50% 0% (0)

Mean score 17.8, median 17.5 · Self-rated confidence: mean 3.1 of 5

n = 27 · mean score 17.78, median 17.5
OptionExpertsShare
Less than 10%933%
10 to 25%1244%
25 to 50%622%
More than 50%00%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=910 to 25% 78%
  • Subtask Bn=11Less than 10% 55%
Q40 Ranked top 3 n = 34

Rank the three tightest bottlenecks for a first 2 Mt/yr green iron export project in a best-in-class location.

Cost of firmed renewable electricity: 50% · 1st 26% Cost of firmed renewable electricity 50% · 1st 26% Offtake certainty: 45% · 1st 26% Offtake certainty 45% · 1st 26% Finance and cost of capital: 39% · 1st 15% Finance and cost of capital 39% · 1st 15% Electrolyser cost and delivery: 23% · 1st 12% Electrolyser cost and delivery 23% · 1st 12% DR-grade ore or beneficiation capacity: 19% · 1st 9% DR-grade ore or beneficiation capacity 19% · 1st 9% Transmission and grid connection: 19% · 1st 12% Transmission and grid connection 19% · 1st 12% Port, HBI handling and shipping: 5% · 1st 0% Port, HBI handling and shipping 5% · 1st 0% Water supply: 1% · 1st 0% Water supply 1% · 1st 0% Permitting, land access and community consent: 0% · 1st 0% Permitting, land access and community consent 0% · 1st 0% Skilled workforce and housing: 0% · 1st 0% Skilled workforce and housing 0% · 1st 0%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 34 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Cost of firmed renewable electricity50%26%74%
Offtake certainty45%26%59%
Finance and cost of capital39%15%62%
Electrolyser cost and delivery23%12%32%
DR-grade ore or beneficiation capacity19%9%29%
Transmission and grid connection19%12%29%
Port, HBI handling and shipping5%0%12%
Water supply1%0%3%
Permitting, land access and community consent0%0%0%
Skilled workforce and housing0%0%0%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9Finance and cost of capital (Borda 56%)
  • Subtask Bn=13Cost of firmed renewable electricity (Borda 51%)
  • Industry and policy referencen=7Offtake certainty (Borda 62%)
Q41 Rating, 1 (Not important) to 5 (Essential) n = 26

How important is public or shared investment in each type of infrastructure for green iron hubs?

High-voltage transmission High-voltage transmission · 3: 20% (n=25)High-voltage transmission · 4: 28% (n=25)High-voltage transmission · 5: 52% (n=25) Mean score (n=25)4.3 Hydrogen pipelines and storage Hydrogen pipelines and storage · 1: 4% (n=24)Hydrogen pipelines and storage · 3: 4% (n=24)Hydrogen pipelines and storage · 4: 46% (n=24)Hydrogen pipelines and storage · 5: 46% (n=24) Mean score (n=24)4.3 Water supply and desalination Water supply and desalination · 2: 8% (n=26)Water supply and desalination · 3: 19% (n=26)Water supply and desalination · 4: 46% (n=26)Water supply and desalination · 5: 27% (n=26) Mean score (n=26)3.9 Port and HBI handling facilities Port and HBI handling facilities · 1: 8% (n=26)Port and HBI handling facilities · 3: 23% (n=26)Port and HBI handling facilities · 4: 54% (n=26)Port and HBI handling facilities · 5: 15% (n=26) Mean score (n=26)3.7 Rail and ore logistics Rail and ore logistics · 1: 4% (n=26)Rail and ore logistics · 2: 23% (n=26)Rail and ore logistics · 3: 12% (n=26)Rail and ore logistics · 4: 50% (n=26)Rail and ore logistics · 5: 12% (n=26) Mean score (n=26)3.4 Common-user industrial precincts Common-user industrial precincts · 1: 12% (n=26)Common-user industrial precincts · 2: 15% (n=26)Common-user industrial precincts · 3: 38% (n=26)Common-user industrial precincts · 4: 19% (n=26)Common-user industrial precincts · 5: 15% (n=26) Mean score (n=26)3.1 Pellet and beneficiation plants Pellet and beneficiation plants · 1: 16% (n=25)Pellet and beneficiation plants · 2: 16% (n=25)Pellet and beneficiation plants · 3: 44% (n=25)Pellet and beneficiation plants · 4: 20% (n=25)Pellet and beneficiation plants · 5: 4% (n=25) Mean score (n=25)2.8 Workforce housing and services Workforce housing and services · 1: 19% (n=26)Workforce housing and services · 2: 23% (n=26)Workforce housing and services · 3: 50% (n=26)Workforce housing and services · 4: 4% (n=26)Workforce housing and services · 5: 4% (n=26) Mean score (n=26)2.5
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
High-voltage transmission254.325.04.0 to 5.0
Hydrogen pipelines and storage244.294.04.0 to 5.0
Water supply and desalination263.924.03.3 to 4.8
Port and HBI handling facilities263.694.03.0 to 4.0
Rail and ore logistics263.424.02.3 to 4.0
Common-user industrial precincts263.123.02.3 to 4.0
Pellet and beneficiation plants252.803.02.0 to 3.0
Workforce housing and services262.503.02.0 to 3.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9Hydrogen pipelines and storage 4.4
  • Subtask Bn=11Hydrogen pipelines and storage 4.6
Q42 Agreement scale n = 33

By 2035, most new hydrogen-based iron capacity will be built in vertically integrated hubs that co-locate ore, renewable electricity, hydrogen and DRI, rather than in disaggregated configurations.

Strongly disagree: 0% (0) Strongly disagree 0% (0) Disagree: 9% (3) Disagree 9% (3) Neutral: 30% (10) Neutral 30% (10) Agree: 42% (14) Agree 42% (14) Strongly agree: 18% (6) Strongly agree 18% (6)

Mean score 0.7, median 1.0

n = 33 · mean score 0.70, median 1.0
OptionExpertsShare
Strongly disagree00%
Disagree39%
Neutral1030%
Agree1442%
Strongly agree618%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9Agree 33%
  • Subtask Bn=12Neutral 42%
  • Industry and policy referencen=7Agree 71%
Q43 Multiple choice, up to 3 n = 30

Which green iron trade corridors will carry the largest volumes in 2040? Select up to three.

Middle East and North Africa to Europe: 60% Middle East and North Africa to Europe 60% Australia to China: 50% Australia to China 50% Australia to Japan and Korea: 43% Australia to Japan and Korea 43% Mostly domestic within China: 40% Mostly domestic within China 40% Brazil to Europe: 27% Brazil to Europe 27% Middle East to South and East Asia: 23% Middle East to South and East Asia 23% Australia to Europe: 17% Australia to Europe 17% Brazil to North-East Asia: 13% Brazil to North-East Asia 13% Mostly domestic within India: 10% Mostly domestic within India 10% Sub-Saharan Africa to Europe: 0% Sub-Saharan Africa to Europe 0% Canada to Europe: 0% Canada to Europe 0%

Share of respondents selecting each option.

n = 30 · each expert could select up to 3
OptionExpertsShare selecting
Middle East and North Africa to Europe1860%
Australia to China1550%
Australia to Japan and Korea1343%
Mostly domestic within China1240%
Brazil to Europe827%
Middle East to South and East Asia723%
Australia to Europe517%
Brazil to North-East Asia413%
Mostly domestic within India310%
Sub-Saharan Africa to Europe00%
Canada to Europe00%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9Middle East and North Africa to Europe 78%
  • Subtask Bn=11Australia to China 64%
  • Industry and policy referencen=6Australia to Japan and Korea 50%
Q44 Low, best and high estimate n = 24

How much hydrogen-based merchant iron (DRI/HBI) will be traded internationally in 2035?

Today's total merchant DRI/HBI trade, almost all natural-gas based, is roughly 20 to 25 million tonnes per year.

Median low

15.0

Median best

35.0

IQR 12.5–60.0

Median high

61.5

060120180240300Interquartile range of best estimates: 12.5–60.0Median best estimate 35.0Expert: low 2, best 4, high 10Expert: low 0, best 5, high 10Expert: low 0, best 5, high 10Expert: low 0, best 5, high 15Expert: low 1, best 5, high 15Expert: low 1, best 5, high 25Expert: low 7, best 15, high 25Expert: low 10, best 15, high 20Expert: low 10, best 20, high 30Expert: low 10, best 25, high 50Expert: low 10, best 30, high 100Expert: low 10, best 35, high 80Expert: low 25, best 35, high 50Expert: low 20, best 36, high 51Expert: low 20, best 43, high 72Expert: low 30, best 50, high 140Expert: low 20, best 60, high 100Expert: low 30, best 60, high 100Expert: low 40, best 60, high 90Expert: low 40, best 60, high 150Expert: low 37, best 78, high 130Expert: low 50, best 100, high 250Expert: low 240, best 280, high 299Expert: low 150, best 300, high 300

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: million tonnes per year of merchant DRI/HBI

Unit: million tonnes per year of merchant DRI/HBI
EstimatenMedianIQRRangeMean
Lowest plausible24156 to 320 to 24031.8
Best estimate243513 to 604 to 30055.5
Highest plausible246224 to 10810 to 30088.4

By Task 52 affiliation (groups of at least 5)

  • Subtask An=7best 50 million tonnes per year of merchant DRI/HBI
  • Subtask Bn=9best 36 million tonnes per year of merchant DRI/HBI
  • Industry and policy referencen=5best 5 million tonnes per year of merchant DRI/HBI
Q45 Rating, 1 (Little influence) to 5 (Decisive) n = 32

How much will each factor drive where new hydrogen-based iron capacity is located over the next decade?

Cost of firmed renewable electricity Cost of firmed renewable electricity · 4: 28% (n=32)Cost of firmed renewable electricity · 5: 72% (n=32) Mean score (n=32)4.7 Political and trade risk Political and trade risk · 3: 27% (n=30)Political and trade risk · 4: 27% (n=30)Political and trade risk · 5: 47% (n=30) Mean score (n=30)4.2 Capital and operating subsidies Capital and operating subsidies · 2: 6% (n=32)Capital and operating subsidies · 3: 13% (n=32)Capital and operating subsidies · 4: 44% (n=32)Capital and operating subsidies · 5: 38% (n=32) Mean score (n=32)4.1 Carbon pricing and border adjustment ex… Carbon pricing and border adjustment exposure · 2: 10% (n=31)Carbon pricing and border adjustment exposure · 3: 23% (n=31)Carbon pricing and border adjustment exposure · 4: 35% (n=31)Carbon pricing and border adjustment exposure · 5: 32% (n=31) Mean score (n=31)3.9 Access to low-cost natural gas as a bri… Access to low-cost natural gas as a bridge · 2: 13% (n=31)Access to low-cost natural gas as a bridge · 3: 26% (n=31)Access to low-cost natural gas as a bridge · 4: 32% (n=31)Access to low-cost natural gas as a bridge · 5: 29% (n=31) Mean score (n=31)3.8 Proximity to green steel demand and off… Proximity to green steel demand and offtakers · 2: 9% (n=32)Proximity to green steel demand and offtakers · 3: 41% (n=32)Proximity to green steel demand and offtakers · 4: 34% (n=32)Proximity to green steel demand and offtakers · 5: 16% (n=32) Mean score (n=32)3.6 Port access and shipping distance to ma… Port access and shipping distance to markets · 2: 10% (n=30)Port access and shipping distance to markets · 3: 40% (n=30)Port access and shipping distance to markets · 4: 37% (n=30)Port access and shipping distance to markets · 5: 13% (n=30) Mean score (n=30)3.5 Water, land and permitting Water, land and permitting · 1: 3% (n=31)Water, land and permitting · 2: 13% (n=31)Water, land and permitting · 3: 29% (n=31)Water, land and permitting · 4: 39% (n=31)Water, land and permitting · 5: 16% (n=31) Mean score (n=31)3.5 Proximity to suitable iron ore Proximity to suitable iron ore · 1: 6% (n=31)Proximity to suitable iron ore · 2: 13% (n=31)Proximity to suitable iron ore · 3: 29% (n=31)Proximity to suitable iron ore · 4: 32% (n=31)Proximity to suitable iron ore · 5: 19% (n=31) Mean score (n=31)3.5 Existing steel plants and skilled workf… Existing steel plants and skilled workforce · 1: 16% (n=31)Existing steel plants and skilled workforce · 2: 16% (n=31)Existing steel plants and skilled workforce · 3: 29% (n=31)Existing steel plants and skilled workforce · 4: 32% (n=31)Existing steel plants and skilled workforce · 5: 6% (n=31) Mean score (n=31)3.0
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Cost of firmed renewable electricity324.725.04.0 to 5.0
Political and trade risk304.204.03.3 to 5.0
Capital and operating subsidies324.134.04.0 to 5.0
Carbon pricing and border adjustment exposure313.904.03.0 to 5.0
Access to low-cost natural gas as a bridge313.774.03.0 to 5.0
Proximity to green steel demand and offtakers323.563.53.0 to 4.0
Port access and shipping distance to markets303.533.53.0 to 4.0
Water, land and permitting313.524.03.0 to 4.0
Proximity to suitable iron ore313.454.03.0 to 4.0
Existing steel plants and skilled workforce312.973.02.0 to 4.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9Cost of firmed renewable electricity 4.8
  • Subtask Bn=12Cost of firmed renewable electricity 4.8
  • Industry and policy referencen=6Cost of firmed renewable electricity 4.5
Q46 Low, best and high estimate n = 16

In 2030, what will be the gap between the DR-grade pellet demand of DRI plants then in commercial operation and the DR-grade pellet supply actually available, merchant plus captive?

Published estimates range from a 16.5 Mt shortfall by 2034 (Midrex) to 133 Mt by 2040 (BloombergNEF); merchant DR-grade supply was roughly 48 Mt in 2025.

Median low

10.0

Median best

25.0

IQR 16.3–53.8

Median high

62.5

0326496128160Interquartile range of best estimates: 16.3–53.8Median best estimate 25.0Expert: low 0, best 0, high 2Expert: low 0, best 2, high 5Expert: low 0, best 5, high 25Expert: low 3, best 5, high 9Expert: low 0, best 20, high 100Expert: low 10, best 20, high 50Expert: low 10, best 20, high 30Expert: low 10, best 20, high 35Expert: low 10, best 30, high 50Expert: low 0, best 40, high 100Expert: low 10, best 50, high 81Expert: low 20, best 50, high 100Expert: low 20, best 65, high 75Expert: low 65, best 65, high 75Expert: low 15, best 75, high 150Expert: low 92, best 100, high 107

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: Mt per year of DR-grade pellets

Unit: Mt per year of DR-grade pellets
EstimatenMedianIQRRangeMean
Lowest plausible16100 to 160 to 9216.6
Best estimate162516 to 540 to 10035.4
Highest plausible166329 to 1002 to 15062.1

By Task 52 affiliation (groups of at least 5)

  • Subtask An=6best 25 Mt per year of DR-grade pellets
  • Subtask Bn=6best 30 Mt per year of DR-grade pellets
Q47 Single choice n = 29

If global high-grade ore supply grew substantially by 2030 (say an additional 100 Mt per year of ore above 65% Fe), how would that affect the availability of DR-grade feed for hydrogen-based ironmaking?

Assume the additional ore still needs beneficiation or pelletising to meet DR-grade specification; high-grade fines are not automatically DR-grade feed.

Materially eases DR-grade feed availability: 24% (7) Materially eases DR-grade feed availability 24% (7) Modest easing: 28% (8) Modest easing 28% (8) Little change: high-grade fines mostly displace blast-furnace feed: 10% (3) Little change: high-grade fines mostly displa… 10% (3) Availability still tightens as DR-grade demand grows faster than qualified supply: 14% (4) Availability still tightens as DR-grade deman… 14% (4) Depends mainly on beneficiation and pelletising capacity: 24% (7) Depends mainly on beneficiation and pelletisi… 24% (7)
n = 29
OptionExpertsShare
Materially eases DR-grade feed availability724%
Modest easing828%
Little change: high-grade fines mostly displace blast-furnace feed310%
Availability still tightens as DR-grade demand grows faster than qualified supply414%
Depends mainly on beneficiation and pelletising capacity724%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9Modest easing 33%
  • Subtask Bn=12Depends mainly on beneficiation and pelletising capacity 33%
Q48 Single choice n = 28

Which ore pathway will supply most of the world's traded green iron tonnage by 2035?

Magnetite beneficiated to DR-grade pellets, fed to shaft-furnace hydrogen DRI: 71% (20) Magnetite beneficiated to DR-grade pellets, f… 71% (20) Hematite fines or lump via an electric smelting furnace, without full beneficiation: 7% (2) Hematite fines or lump via an electric smelti… 7% (2) Roughly even split between the two: 11% (3) Roughly even split between the two 11% (3) Neither; global green iron trade stays below 5 Mt per year in 2035: 11% (3) Neither; global green iron trade stays below … 11% (3)
n = 28
OptionExpertsShare
Magnetite beneficiated to DR-grade pellets, fed to shaft-furnace hydrogen DRI2071%
Hematite fines or lump via an electric smelting furnace, without full beneficiation27%
Roughly even split between the two311%
Neither; global green iron trade stays below 5 Mt per year in 2035311%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9Magnetite beneficiated to DR-grade pellets, fed to shaft-furnace hydrogen DRI 67%
  • Subtask Bn=11Magnetite beneficiated to DR-grade pellets, fed to shaft-furnace hydrogen DRI 64%
Q49 Probability, 0 to 100% n = 27

What is the probability that by 2035 more than half of global DRI output is made from ore that is below 65% Fe as mined, before beneficiation, whether the ore is upgraded to DR grade by beneficiation or used at lower grade with an electric smelter downstream?

0% means certain not to, 100% means certain. Today almost all DRI uses reactor feed of 67% Fe or more, most of it from ore that was already high grade as mined.

0%20%40%60%80%100%More than half of global DRI output from ore below 65% Fe by 2035: median 40%, IQR 23–58 (n=27) More than half of global DRI output from or… 40%

Dot: median · bar: interquartile range

Probability sliders, 0 to 100% · rows sorted by mean
RownMedian %IQR %Range %
More than half of global DRI output from ore below 65% Fe by 2035274023 to 585 to 95

By Task 52 affiliation (groups of at least 5)

  • Subtask An=9More than half of global DRI output from ore below 65% Fe by 2035 35%
  • Subtask Bn=10More than half of global DRI output from ore below 65% Fe by 2035 50%
  • Industry and policy referencen=5More than half of global DRI output from ore below 65% Fe by 2035 20%
Q50 Low, best and high estimate n = 17

For feed charged to a shaft furnace as DR pellets or lump ore, below what Fe grade of that feed can the plant no longer operate commercially without further upgrading?

Give the Fe grade of the feed as charged (pellets or lump), not of the ore as mined. Lowest and highest plausible first, then your best estimate.

Median low

61.0

Median best

65.0

IQR 62.0–65.0

Median high

67.0

505458626670Interquartile range of best estimates: 62.0–65.0Median best estimate 65.0Expert: low 50, best 60, high 70Expert: low 50, best 60, high 70Expert: low 55, best 60, high 65Expert: low 58, best 62, high 67.5Expert: low 60, best 62, high 65Expert: low 60, best 62, high 64.5Expert: low 61, best 63, high 65Expert: low 61.5, best 63.5, high 64.5Expert: low 55, best 65, high 67Expert: low 55, best 65, high 67Expert: low 62, best 65, high 68Expert: low 62, best 65, high 68Expert: low 63, best 65, high 67Expert: low 64, best 65, high 67Expert: low 65, best 65, high 65Expert: low 62, best 66, high 68Expert: low 66, best 67, high 68

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: % Fe

Unit: % Fe
EstimatenMedianIQRRangeMean
Lowest plausible1761.055.0 to 62.050.0 to 66.059.38
Best estimate1765.062.0 to 65.060.0 to 67.063.56
Highest plausible1767.065.0 to 68.064.5 to 70.066.85

By Task 52 affiliation (groups of at least 5)

  • Subtask An=8best 62.8% Fe
  • Subtask Bn=5best 65.0% Fe
Q51 Low, best and high estimate n = 18

What is the cost premium for producing DR-grade pellets, in US dollars per tonne of pellets?

Per dry tonne of pellets produced. Include grinding, beneficiation to DR-grade chemistry and pelletising; exclude mining and transport. Compare with producing sinter feed or fines from the same ore at a best-in-class location.

Median low

20.0

Median best

35.0

IQR 28.5–50.0

Median high

52.5

020406080100Interquartile range of best estimates: 28.5–50.0Median best estimate 35.0Expert: low 0, best 15, high 25Expert: low 10, best 20, high 35Expert: low 18, best 20, high 40Expert: low 15, best 25, high 40Expert: low 20, best 28, high 35Expert: low 20, best 30, high 50Expert: low 25, best 30, high 50Expert: low 15, best 35, high 70Expert: low 20, best 35, high 55Expert: low 26, best 35, high 40Expert: low 20, best 36, high 50Expert: low 30, best 40, high 60Expert: low 20, best 50, high 80Expert: low 25, best 50, high 75Expert: low 30, best 50, high 60Expert: low 40, best 60, high 75Expert: low 50, best 60, high 100Expert: low 50, best 80, high 100

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: USD per tonne of pellets

Unit: USD per tonne of pellets
EstimatenMedianIQRRangeMean
Lowest plausible182019 to 290 to 5024.1
Best estimate183529 to 5015 to 8038.8
Highest plausible185340 to 7425 to 10057.8

By Task 52 affiliation (groups of at least 5)

  • Subtask An=6best 35 USD per tonne of pellets
  • Subtask Bn=6best 33 USD per tonne of pellets
Q52 Rating, 1 (No constraint) to 5 (Severe constraint) n = 25

How serious a constraint will each of these be on global hydrogen-based ironmaking and EAF capacity additions between 2026 and 2032?

Electrolyser manufacturing capacity and… Electrolyser manufacturing capacity and delivery times · 2: 13% (n=24)Electrolyser manufacturing capacity and delivery times · 3: 42% (n=24)Electrolyser manufacturing capacity and delivery times · 4: 17% (n=24)Electrolyser manufacturing capacity and delivery times · 5: 29% (n=24) Mean score (n=24)3.6 High-voltage transformers and grid equi… High-voltage transformers and grid equipment · 2: 17% (n=24)High-voltage transformers and grid equipment · 3: 29% (n=24)High-voltage transformers and grid equipment · 4: 29% (n=24)High-voltage transformers and grid equipment · 5: 25% (n=24) Mean score (n=24)3.6 Shaft-furnace and DRI-plant EPC and equ… Shaft-furnace and DRI-plant EPC and equipment capacity · 1: 8% (n=25)Shaft-furnace and DRI-plant EPC and equipment capacity · 2: 16% (n=25)Shaft-furnace and DRI-plant EPC and equipment capacity · 3: 16% (n=25)Shaft-furnace and DRI-plant EPC and equipment capacity · 4: 40% (n=25)Shaft-furnace and DRI-plant EPC and equipment capacity · 5: 20% (n=25) Mean score (n=25)3.5 Electric smelting furnace equipment ava… Electric smelting furnace equipment availability · 1: 8% (n=25)Electric smelting furnace equipment availability · 2: 8% (n=25)Electric smelting furnace equipment availability · 3: 40% (n=25)Electric smelting furnace equipment availability · 4: 28% (n=25)Electric smelting furnace equipment availability · 5: 16% (n=25) Mean score (n=25)3.4 Specialist engineering and construction… Specialist engineering and construction labour · 1: 8% (n=25)Specialist engineering and construction labour · 2: 12% (n=25)Specialist engineering and construction labour · 3: 44% (n=25)Specialist engineering and construction labour · 4: 28% (n=25)Specialist engineering and construction labour · 5: 8% (n=25) Mean score (n=25)3.2 EAF graphite electrode and needle-coke … EAF graphite electrode and needle-coke supply · 1: 9% (n=22)EAF graphite electrode and needle-coke supply · 2: 23% (n=22)EAF graphite electrode and needle-coke supply · 3: 36% (n=22)EAF graphite electrode and needle-coke supply · 4: 23% (n=22)EAF graphite electrode and needle-coke supply · 5: 9% (n=22) Mean score (n=22)3.0
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Electrolyser manufacturing capacity and delivery times243.633.03.0 to 5.0
High-voltage transformers and grid equipment243.634.03.0 to 4.3
Shaft-furnace and DRI-plant EPC and equipment capacity253.484.03.0 to 4.0
Electric smelting furnace equipment availability253.363.03.0 to 4.0
Specialist engineering and construction labour253.163.03.0 to 4.0
EAF graphite electrode and needle-coke supply223.003.02.0 to 4.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=8Shaft-furnace and DRI-plant EPC and equipment capacity 3.9
  • Subtask Bn=8High-voltage transformers and grid equipment 3.9
Q53 Year percentiles n = 22

In which year will a liquid merchant market for hydrogen-based HBI exist, with multiple independent buyers and sellers and cargoes routinely traded against a published price benchmark?

Comparable in market structure, not volume, to today's seaborne iron ore market. Fill the 10% and 90% years first.

10% chance by

2034

IQR 2030–2035

50% chance by

2040

IQR 2037–2042

90% chance by

2048

IQR 2041–2050

0%25%50%75%100%20272031203520392043204720502027: 1% pooled probability2028: 2% pooled probability2029: 3% pooled probability2030: 5% pooled probability2031: 9% pooled probability2032: 13% pooled probability2033: 16% pooled probability2034: 21% pooled probability2035: 25% pooled probability2036: 32% pooled probability2037: 38% pooled probability2038: 44% pooled probability2039: 50% pooled probability2040: 57% pooled probability2041: 63% pooled probability2042: 67% pooled probability2043: 70% pooled probability2044: 73% pooled probability2045: 77% pooled probability2046: 79% pooled probability2047: 82% pooled probability2048: 84% pooled probability2049: 86% pooled probability2050: 88% pooled probability

Pooled cumulative probability: average of each expert's piecewise-linear curve through their 10%, 50% and 90% years.

“After 2050” is counted as 2055 for the statistics
PercentilenMedian yearIQRRange“Never”
10% chance by2220342030 to 20352028 to 20550
50% chance by2220402037 to 20422032 to 20550
90% chance by2220482041 to 20502035 to 20550

By Task 52 affiliation (groups of at least 5)

  • Subtask An=850% chance by 2040
  • Subtask Bn=650% chance by 2039
  • Industry and policy referencen=550% chance by 2040
Projects, finance and offtake 26 experts · 6 questions
Q54 Low, best and high estimate n = 12

What cost of debt will a first-of-a-kind 2 Mt per year hydrogen DRI project with no operating track record face at final investment decision in 2026 or 2027, without government guarantees?

BCG analysis cited in 2026 reporting found green hydrogen project debt costs more than three times that of mature renewables.

Median low

6.5

Median best

9.0

IQR 8.0–12.6

Median high

15.0

25.69.212.816.420Interquartile range of best estimates: 8.0–12.6Median best estimate 9.0Expert: low 3, best 3, high 4Expert: low 6, best 7, high 9Expert: low 3, best 8, high 15Expert: low 5, best 8, high 15Expert: low 3, best 8.25, high 15.25Expert: low 6, best 9, high 14Expert: low 7, best 9, high 12Expert: low 8, best 10, high 12Expert: low 9.25, best 12.25, high 17.25Expert: low 8, best 13.75, high 15Expert: low 10, best 15, high 20Expert: low 10, best 20, high 20

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: % per year, cost of debt

Unit: % per year, cost of debt
EstimatenMedianIQRRangeMean
Lowest plausible126.54.5 to 8.33.0 to 10.06.52
Best estimate129.08.0 to 12.63.0 to 20.010.27
Highest plausible1215.012.0 to 15.84.0 to 20.014.04
Q55 Ranked top 3 n = 21

Rank the three financing or risk-sharing mechanisms that would most increase the number of hydrogen-based iron and steel projects reaching final investment decision by 2029.

Government offtake or revenue-support contracts (carbon contracts for difference): 76% · 1st 62% Government offtake or revenue-support contrac… 76% · 1st 62% Buyer-backed offtake agreements usable as loan collateral: 43% · 1st 19% Buyer-backed offtake agreements usable as loa… 43% · 1st 19% Direct capital grants or public equity co-investment: 33% · 1st 14% Direct capital grants or public equity co-inv… 33% · 1st 14% Sovereign or strategic-investor equity stakes: 17% · 1st 5% Sovereign or strategic-investor equity stakes 17% · 1st 5% Multilateral development bank guarantees or first-loss capital: 8% · 1st 0% Multilateral development bank guarantees or f… 8% · 1st 0% Blended or concessional finance from climate funds: 8% · 1st 0% Blended or concessional finance from climate … 8% · 1st 0% Export credit agency-backed debt: 6% · 1st 0% Export credit agency-backed debt 6% · 1st 0% Insurance covering technology or completion risk: 6% · 1st 0% Insurance covering technology or completion r… 6% · 1st 0% Green or sustainability-linked bond markets: 2% · 1st 0% Green or sustainability-linked bond markets 2% · 1st 0%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 21 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Government offtake or revenue-support contracts (carbon contracts for difference)76%62%86%
Buyer-backed offtake agreements usable as loan collateral43%19%62%
Direct capital grants or public equity co-investment33%14%52%
Sovereign or strategic-investor equity stakes17%5%33%
Multilateral development bank guarantees or first-loss capital8%0%19%
Blended or concessional finance from climate funds8%0%14%
Export credit agency-backed debt6%0%14%
Insurance covering technology or completion risk6%0%14%
Green or sustainability-linked bond markets2%0%5%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5Government offtake or revenue-support contracts (carbon contracts for difference) (Borda 60%)
  • Subtask Bn=5Government offtake or revenue-support contracts (carbon contracts for difference) (Borda 80%)
  • Industry and policy referencen=6Government offtake or revenue-support contracts (carbon contracts for difference) (Borda 72%)
Q56 Choice grid n = 15

In a typical hydrogen DRI project financing today, which party bears most of each risk?

Offtake and demand risk Offtake and demand risk · Project sponsor or equity: 53% (n=15)Offtake and demand risk · Government support scheme: 7% (n=15)Offtake and demand risk · Buyer, via offtake terms: 33% (n=15)Offtake and demand risk · Lender or debt: 7% (n=15) Mean score (n=15)n=15 Hydrogen and electricity price risk Hydrogen and electricity price risk · Project sponsor or equity: 60% (n=15)Hydrogen and electricity price risk · Government support scheme: 7% (n=15)Hydrogen and electricity price risk · Buyer, via offtake terms: 7% (n=15)Hydrogen and electricity price risk · Largely unallocated or uninsured: 27% (n=15) Mean score (n=15)n=15 Technology and completion risk Technology and completion risk · Project sponsor or equity: 71% (n=14)Technology and completion risk · Government support scheme: 14% (n=14)Technology and completion risk · Lender or debt: 7% (n=14)Technology and completion risk · Largely unallocated or uninsured: 7% (n=14) Mean score (n=14)n=14 Policy durability risk, such as subsidy… Policy durability risk, such as subsidy or contract survival · Project sponsor or equity: 47% (n=15)Policy durability risk, such as subsidy or contract survival · Government support scheme: 27% (n=15)Policy durability risk, such as subsidy or contract survival · Buyer, via offtake terms: 7% (n=15)Policy durability risk, such as subsidy or contract survival · Largely unallocated or uninsured: 20% (n=15) Mean score (n=15)n=15 Political and cross-border trade risk Political and cross-border trade risk · Project sponsor or equity: 40% (n=15)Political and cross-border trade risk · Government support scheme: 27% (n=15)Political and cross-border trade risk · Buyer, via offtake terms: 13% (n=15)Political and cross-border trade risk · Largely unallocated or uninsured: 20% (n=15) Mean score (n=15)n=15
Project sponsor or equityGovernment support schemeBuyer, via offtake termsLender or debtLargely unallocated or uninsuredNumber at right: mean score
Share of the experts answering each row
RownProject sponsor or equityGovernment support schemeBuyer, via offtake termsLender or debtLargely unallocated or uninsured
Offtake and demand risk1553%7%33%7%0%
Hydrogen and electricity price risk1560%7%7%0%27%
Technology and completion risk1471%14%0%7%7%
Policy durability risk, such as subsidy or contract survival1547%27%7%0%20%
Political and cross-border trade risk1540%27%13%0%20%

By Task 52 affiliation (groups of at least 5)

  • Industry and policy referencen=5Offtake and demand risk: Project sponsor or equity 40%
Q57 Choice grid n = 13

Based on dated, publicly reported assessments, what premium do you judge is typical in 2026 for near-zero or low-emission steel, by buyer segment?

Use published assessments only (for example, Fastmarkets' published European green flat-steel premium assessment of roughly EUR 125 per tonne in early 2026). Do not use your own or a counterparty's contract terms.

Automotive OEM long-term contracts Automotive OEM long-term contracts · No premium: 8% (n=13)Automotive OEM long-term contracts · Up to USD 50/t: 15% (n=13)Automotive OEM long-term contracts · USD 50 to 120/t: 31% (n=13)Automotive OEM long-term contracts · USD 120 to 200/t: 46% (n=13) Mean score (n=13)103.8 Construction and infrastructure Construction and infrastructure · No premium: 46% (n=13)Construction and infrastructure · Up to USD 50/t: 38% (n=13)Construction and infrastructure · USD 50 to 120/t: 8% (n=13)Construction and infrastructure · USD 120 to 200/t: 8% (n=13) Mean score (n=13)28.5 Wind and renewable-equipment manufactur… Wind and renewable-equipment manufacturers · No premium: 30% (n=10)Wind and renewable-equipment manufacturers · Up to USD 50/t: 20% (n=10)Wind and renewable-equipment manufacturers · USD 50 to 120/t: 30% (n=10)Wind and renewable-equipment manufacturers · USD 120 to 200/t: 20% (n=10) Mean score (n=10)62.5 Appliance and consumer goods Appliance and consumer goods · No premium: 18% (n=11)Appliance and consumer goods · Up to USD 50/t: 36% (n=11)Appliance and consumer goods · USD 50 to 120/t: 27% (n=11)Appliance and consumer goods · USD 120 to 200/t: 18% (n=11) Mean score (n=11)61.4 Spot market, service centres and stockh… Spot market, service centres and stockholders · No premium: 55% (n=11)Spot market, service centres and stockholders · Up to USD 50/t: 18% (n=11)Spot market, service centres and stockholders · USD 50 to 120/t: 18% (n=11)Spot market, service centres and stockholders · USD 120 to 200/t: 9% (n=11) Mean score (n=11)34.5
No premiumUp to USD 50/tUSD 50 to 120/tUSD 120 to 200/tMore than USD 200/tNumber at right: mean score
Share of the experts answering each row
RownNo premiumUp to USD 50/tUSD 50 to 120/tUSD 120 to 200/tMore than USD 200/t
Automotive OEM long-term contracts138%15%31%46%0%
Construction and infrastructure1346%38%8%8%0%
Wind and renewable-equipment manufacturers1030%20%30%20%0%
Appliance and consumer goods1118%36%27%18%0%
Spot market, service centres and stockholders1155%18%18%9%0%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5Automotive OEM long-term contracts: USD 50 to 120/t 60%
Q58 Ranked top 3 n = 15

By 2030, which contract structure will be most common for large-volume green iron and steel offtake agreements of 250,000 tonnes per year or more? Rank the three you expect to be most common, most common first.

Fixed premium over a benchmark index for the contract life: 53% · 1st 33% Fixed premium over a benchmark index for the … 53% · 1st 33% Government contract for difference covers the gap; buyer pays the reference price: 40% · 1st 13% Government contract for difference covers the… 40% · 1st 13% No standard structure; deals stay bespoke: 33% · 1st 27% No standard structure; deals stay bespoke 33% · 1st 27% Cost-plus pass-through of the seller's input costs: 29% · 1st 13% Cost-plus pass-through of the seller's input … 29% · 1st 13% Premium indexed to a published green-steel price assessment: 27% · 1st 13% Premium indexed to a published green-steel pr… 27% · 1st 13% Floor and ceiling collar around an index: 18% · 1st 0% Floor and ceiling collar around an index 18% · 1st 0%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 15 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Fixed premium over a benchmark index for the contract life53%33%67%
Government contract for difference covers the gap; buyer pays the reference price40%13%60%
No standard structure; deals stay bespoke33%27%47%
Cost-plus pass-through of the seller's input costs29%13%47%
Premium indexed to a published green-steel price assessment27%13%40%
Floor and ceiling collar around an index18%0%40%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5Fixed premium over a benchmark index for the contract life (Borda 53%)
Q59 Single choice n = 19

Of new hydrogen-based DRI capacity reaching commercial operation worldwide between 2026 and 2035, what share will be brownfield conversions of existing integrated steelworks rather than greenfield builds?

Less than 20% brownfield; greenfield dominates: 32% (6) Less than 20% brownfield; greenfield dominates 32% (6) 20 to 40% brownfield: 32% (6) 20 to 40% brownfield 32% (6) 40 to 60%, roughly even: 32% (6) 40 to 60%, roughly even 32% (6) More than 60% brownfield; conversion dominates: 5% (1) More than 60% brownfield; conversion dominates 5% (1)

Mean score 32.4, median 30.0

n = 19 · mean score 32.37, median 30.0
OptionExpertsShare
Less than 20% brownfield; greenfield dominates632%
20 to 40% brownfield632%
40 to 60%, roughly even632%
More than 60% brownfield; conversion dominates15%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=540 to 60%, roughly even 40%
  • Industry and policy referencen=6Less than 20% brownfield; greenfield dominates 33%
Pathways, competitiveness and power systems 42 experts · 16 questions
Q60 Choice grid n = 32

What production-cost premium will domestic hydrogen-based DRI-EAF crude steel carry over BF-BOF in Europe and in China, excluding carbon pricing and subsidies?

Europe, 2030 Europe, 2030 · 0 to 10%: 9% (n=32)Europe, 2030 · 10 to 25%: 6% (n=32)Europe, 2030 · 25 to 50%: 47% (n=32)Europe, 2030 · Above 50%: 38% (n=32) Mean score (n=32)41.6 Europe, 2035 Europe, 2035 · 0 to 10%: 9% (n=32)Europe, 2035 · 10 to 25%: 31% (n=32)Europe, 2035 · 25 to 50%: 44% (n=32)Europe, 2035 · Above 50%: 16% (n=32) Mean score (n=32)31.7 China, 2030 China, 2030 · 0 to 10%: 7% (n=29)China, 2030 · 10 to 25%: 31% (n=29)China, 2030 · 25 to 50%: 52% (n=29)China, 2030 · Above 50%: 10% (n=29) Mean score (n=29)31.4 China, 2035 China, 2035 · 0 to 10%: 24% (n=29)China, 2035 · 10 to 25%: 45% (n=29)China, 2035 · 25 to 50%: 28% (n=29)China, 2035 · Above 50%: 3% (n=29) Mean score (n=29)21.5
Below BF-BOF cost0 to 10%10 to 25%25 to 50%Above 50%Number at right: mean score
Share of the experts answering each row
RownBelow BF-BOF cost0 to 10%10 to 25%25 to 50%Above 50%
Europe, 2030320%9%6%47%38%
Europe, 2035320%9%31%44%16%
China, 2030290%7%31%52%10%
China, 2035290%24%45%28%3%

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=5Europe, 2030: 25 to 50% 60%
  • Subtask Cn=12Europe, 2030: Above 50% 58%
  • Linkage and exchangen=7Europe, 2030: 25 to 50% 86%
Q61 Low, best and high estimate n = 30

What carbon price would make hydrogen-based DRI-EAF steel cost-competitive with unabated BF-BOF in Europe in 2030, without other subsidies?

Median low

110.0

Median best

165.0

IQR 141.3–200.0

Median high

232.5

080160240320400Interquartile range of best estimates: 141.3–200.0Median best estimate 165.0Expert: low 75, best 90, high 205Expert: low 75, best 100, high 150Expert: low 80, best 100, high 150Expert: low 60, best 120, high 250Expert: low 75, best 120, high 165Expert: low 90, best 140, high 220Expert: low 100, best 140, high 180Expert: low 110, best 140, high 200Expert: low 90, best 145, high 240Expert: low 90, best 150, high 200Expert: low 100, best 150, high 200Expert: low 100, best 150, high 200Expert: low 100, best 150, high 200Expert: low 110, best 160, high 180Expert: low 150, best 160, high 200Expert: low 100, best 170, high 200Expert: low 120, best 180, high 250Expert: low 165, best 190, high 225Expert: low 120, best 200, high 300Expert: low 120, best 200, high 275Expert: low 150, best 200, high 250Expert: low 150, best 200, high 350Expert: low 150, best 200, high 300Expert: low 180, best 200, high 250Expert: low 120, best 205, high 320Expert: low 100, best 220, high 295Expert: low 180, best 230, high 250Expert: low 130, best 250, high 400Expert: low 150, best 250, high 300Expert: low 200, best 250, high 300

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: USD per tonne CO2

Unit: USD per tonne CO2
EstimatenMedianIQRRangeMean
Lowest plausible3011093 to 15060 to 200118.0
Best estimate30165141 to 20090 to 250172.0
Highest plausible30233200 to 290150 to 400240.2

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=12best 185 USD per tonne CO2
  • Linkage and exchangen=6best 130 USD per tonne CO2
Q62 Low, best and high estimate n = 34

What will low-emission hydrogen cost, delivered to a DRI plant in a best-in-class location, in 2035?

Include production, storage and buffering needed for continuous DRI operation.

Median low

2.0

Median best

3.0

IQR 2.5–4.4

Median high

4.8

0246810Interquartile range of best estimates: 2.5–4.4Median best estimate 3.0Expert: low 1, best 1.2, high 1.5Expert: low 1.1, best 1.7, high 3.5Expert: low 1, best 2, high 3.2Expert: low 1.5, best 2, high 2.5Expert: low 1.8, best 2, high 2.3Expert: low 2, best 2.4, high 3Expert: low 2, best 2.5, high 5Expert: low 2, best 2.5, high 3.5Expert: low 2, best 2.5, high 4Expert: low 2, best 2.5, high 5Expert: low 1.3, best 2.8, high 5Expert: low 2, best 2.8, high 4.5Expert: low 2, best 2.8, high 3.9Expert: low 1, best 3, high 4Expert: low 1.5, best 3, high 5.5Expert: low 1.5, best 3, high 4Expert: low 2, best 3, high 3Expert: low 2, best 3, high 7Expert: low 2, best 3, high 5Expert: low 2.5, best 3, high 4Expert: low 2.5, best 3.5, high 4.5Expert: low 2.8, best 3.5, high 4.5Expert: low 2.8, best 3.5, high 4.2Expert: low 1, best 4, high 8Expert: low 2, best 4, high 5.5Expert: low 4, best 4.5, high 6Expert: low 2.4, best 4.8, high 6.9Expert: low 3, best 5, high 6Expert: low 4, best 6, high 7Expert: low 4, best 6, high 10Expert: low 6.5, best 8.5, high 10Expert: low 5, best 10, high 10Expert: low 8, best 10, high 10Expert: low 10, best 10, high 10

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: USD per kg hydrogen

Unit: USD per kg hydrogen
EstimatenMedianIQRRangeMean
Lowest plausible342.01.6 to 2.81.0 to 10.02.71
Best estimate343.02.5 to 4.41.2 to 10.03.94
Highest plausible344.83.9 to 6.71.5 to 10.05.35

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=5best 3.0 USD per kg hydrogen
  • Subtask Cn=12best 2.9 USD per kg hydrogen
  • Linkage and exchangen=7best 2.8 USD per kg hydrogen
  • Industry and policy referencen=5best 6.0 USD per kg hydrogen
Q63 Probability, 0 to 100% n = 34

What is the probability that each milestone is reached by the end of 2030?

0% means impossible, 100% means certain.

0%20%40%60%80%100%China has three or more hydrogen shaft furnaces of 1 Mt/yr or larger operating: median 70%, IQR 50–80 (n=29) China has three or more hydrogen shaft furn… 70%EU CBAM applies to iron and steel at full strength without major derogations: median 45%, IQR 25–70 (n=30) EU CBAM applies to iron and steel at full s… 45%A near-zero steel definition is mutually recognised by the EU, Japan and Korea: median 45%, IQR 18–60 (n=28) A near-zero steel definition is mutually re… 45%At least 10 Mt/yr of hydrogen-based DRI capacity operating globally: median 25%, IQR 11–50 (n=34) At least 10 Mt/yr of hydrogen-based DRI cap… 25%A commercial-scale hydrogen-based DRI plant is operating in North America: median 20%, IQR 10–31 (n=28) A commercial-scale hydrogen-based DRI plant… 20%At least 5 Mt/yr of green iron traded between continents: median 25%, IQR 10–35 (n=33) At least 5 Mt/yr of green iron traded betwe… 25%

Dot: median · bar: interquartile range

Probability sliders, 0 to 100% · rows sorted by mean
RownMedian %IQR %Range %
China has three or more hydrogen shaft furnaces of 1 Mt/yr or larger operating297050 to 805 to 100
EU CBAM applies to iron and steel at full strength without major derogations304525 to 700 to 100
A near-zero steel definition is mutually recognised by the EU, Japan and Korea284518 to 605 to 80
At least 10 Mt/yr of hydrogen-based DRI capacity operating globally342511 to 500 to 85
A commercial-scale hydrogen-based DRI plant is operating in North America282010 to 310 to 80
At least 5 Mt/yr of green iron traded between continents332510 to 350 to 100

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=12China has three or more hydrogen shaft furnaces of 1 Mt/yr or larger operating 78%
  • Linkage and exchangen=7China has three or more hydrogen shaft furnaces of 1 Mt/yr or larger operating 60%
  • Industry and policy referencen=5China has three or more hydrogen shaft furnaces of 1 Mt/yr or larger operating 63%
Q64 Single choice n = 33

What share of Europe's primary iron units for near-zero steel will be imported as HBI or DRI in 2040, rather than reduced in Europe?

Less than 10%: 6% (2) Less than 10% 6% (2) 10 to 25%: 39% (13) 10 to 25% 39% (13) 25 to 50%: 39% (13) 25 to 50% 39% (13) 50 to 75%: 9% (3) 50 to 75% 9% (3) More than 75%: 6% (2) More than 75% 6% (2)

Mean score 32.8, median 37.5 · Self-rated confidence: mean 3.0 of 5

n = 33 · mean score 32.80, median 37.5
OptionExpertsShare
Less than 10%26%
10 to 25%1339%
25 to 50%1339%
50 to 75%39%
More than 75%26%

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=525 to 50% 60%
  • Subtask Cn=1225 to 50% 50%
  • Linkage and exchangen=610 to 25% 50%
  • Industry and policy referencen=510 to 25% 40%
Q65 Rating, 1 (Minor) to 5 (Critical) n = 32

How important is each power-system issue for hydrogen-based ironmaking at multi-million-tonne scale?

Cost of firming variable renewables Cost of firming variable renewables · 2: 3% (n=30)Cost of firming variable renewables · 3: 13% (n=30)Cost of firming variable renewables · 4: 33% (n=30)Cost of firming variable renewables · 5: 50% (n=30) Mean score (n=30)4.3 Electricity market design and industria… Electricity market design and industrial tariffs · 1: 3% (n=29)Electricity market design and industrial tariffs · 2: 3% (n=29)Electricity market design and industrial tariffs · 3: 14% (n=29)Electricity market design and industrial tariffs · 4: 34% (n=29)Electricity market design and industrial tariffs · 5: 45% (n=29) Mean score (n=29)4.1 Grid connection lead times Grid connection lead times · 1: 3% (n=29)Grid connection lead times · 2: 3% (n=29)Grid connection lead times · 3: 34% (n=29)Grid connection lead times · 4: 17% (n=29)Grid connection lead times · 5: 41% (n=29) Mean score (n=29)3.9 Large-scale hydrogen storage Large-scale hydrogen storage · 2: 13% (n=32)Large-scale hydrogen storage · 3: 22% (n=32)Large-scale hydrogen storage · 4: 41% (n=32)Large-scale hydrogen storage · 5: 25% (n=32) Mean score (n=32)3.8 Transmission build-out Transmission build-out · 1: 3% (n=31)Transmission build-out · 2: 6% (n=31)Transmission build-out · 3: 26% (n=31)Transmission build-out · 4: 39% (n=31)Transmission build-out · 5: 26% (n=31) Mean score (n=31)3.8 Flexible electrolyser operation Flexible electrolyser operation · 2: 19% (n=32)Flexible electrolyser operation · 3: 19% (n=32)Flexible electrolyser operation · 4: 34% (n=32)Flexible electrolyser operation · 5: 28% (n=32) Mean score (n=32)3.7 DRI plant turndown and ramping DRI plant turndown and ramping · 1: 6% (n=31)DRI plant turndown and ramping · 2: 13% (n=31)DRI plant turndown and ramping · 3: 52% (n=31)DRI plant turndown and ramping · 4: 29% (n=31) Mean score (n=31)3.0 Value of DRI plants as flexible demand … Value of DRI plants as flexible demand for the grid · 1: 7% (n=29)Value of DRI plants as flexible demand for the grid · 2: 24% (n=29)Value of DRI plants as flexible demand for the grid · 3: 38% (n=29)Value of DRI plants as flexible demand for the grid · 4: 28% (n=29)Value of DRI plants as flexible demand for the grid · 5: 3% (n=29) Mean score (n=29)3.0
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Cost of firming variable renewables304.304.54.0 to 5.0
Electricity market design and industrial tariffs294.144.04.0 to 5.0
Grid connection lead times293.904.03.0 to 5.0
Large-scale hydrogen storage323.784.03.0 to 4.3
Transmission build-out313.774.03.0 to 4.5
Flexible electrolyser operation323.724.03.0 to 5.0
DRI plant turndown and ramping313.033.03.0 to 4.0
Value of DRI plants as flexible demand for the grid292.973.02.0 to 4.0

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=12Electricity market design and industrial tariffs 4.1
  • Linkage and exchangen=5Electricity market design and industrial tariffs 4.2
  • Industry and policy referencen=6Cost of firming variable renewables 4.6
Q66 Rating, 1 (Ineffective) to 5 (Highly effective) n = 33

How effective will each policy instrument be at sustaining hydrogen-based steelmaking after 2030, once first plants are built?

Domestic carbon pricing Domestic carbon pricing · 1: 3% (n=32)Domestic carbon pricing · 2: 9% (n=32)Domestic carbon pricing · 3: 13% (n=32)Domestic carbon pricing · 4: 28% (n=32)Domestic carbon pricing · 5: 47% (n=32) Mean score (n=32)4.1 Border carbon adjustment Border carbon adjustment · 2: 6% (n=32)Border carbon adjustment · 3: 9% (n=32)Border carbon adjustment · 4: 56% (n=32)Border carbon adjustment · 5: 28% (n=32) Mean score (n=32)4.1 Carbon contracts for difference Carbon contracts for difference · 2: 7% (n=30)Carbon contracts for difference · 3: 20% (n=30)Carbon contracts for difference · 4: 43% (n=30)Carbon contracts for difference · 5: 30% (n=30) Mean score (n=30)4.0 Product standards and embodied-carbon l… Product standards and embodied-carbon limits · 2: 3% (n=32)Product standards and embodied-carbon limits · 3: 31% (n=32)Product standards and embodied-carbon limits · 4: 53% (n=32)Product standards and embodied-carbon limits · 5: 13% (n=32) Mean score (n=32)3.8 Lead markets and quotas Lead markets and quotas · 2: 3% (n=32)Lead markets and quotas · 3: 44% (n=32)Lead markets and quotas · 4: 31% (n=32)Lead markets and quotas · 5: 22% (n=32) Mean score (n=32)3.7 Hydrogen subsidies Hydrogen subsidies · 2: 18% (n=33)Hydrogen subsidies · 3: 21% (n=33)Hydrogen subsidies · 4: 36% (n=33)Hydrogen subsidies · 5: 24% (n=33) Mean score (n=33)3.7 Green iron trade agreements Green iron trade agreements · 2: 18% (n=33)Green iron trade agreements · 3: 27% (n=33)Green iron trade agreements · 4: 30% (n=33)Green iron trade agreements · 5: 24% (n=33) Mean score (n=33)3.6 Green public procurement Green public procurement · 2: 16% (n=32)Green public procurement · 3: 38% (n=32)Green public procurement · 4: 25% (n=32)Green public procurement · 5: 22% (n=32) Mean score (n=32)3.5 Plurilateral clubs and sectoral agreeme… Plurilateral clubs and sectoral agreements · 1: 6% (n=32)Plurilateral clubs and sectoral agreements · 2: 13% (n=32)Plurilateral clubs and sectoral agreements · 3: 44% (n=32)Plurilateral clubs and sectoral agreements · 4: 31% (n=32)Plurilateral clubs and sectoral agreements · 5: 6% (n=32) Mean score (n=32)3.2
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Domestic carbon pricing324.064.03.8 to 5.0
Border carbon adjustment324.064.04.0 to 5.0
Carbon contracts for difference303.974.03.3 to 5.0
Product standards and embodied-carbon limits323.754.03.0 to 4.0
Lead markets and quotas323.724.03.0 to 4.0
Hydrogen subsidies333.674.03.0 to 4.0
Green iron trade agreements333.614.03.0 to 4.0
Green public procurement323.533.03.0 to 4.0
Plurilateral clubs and sectoral agreements323.193.03.0 to 4.0

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=5Carbon contracts for difference 4.5
  • Subtask Cn=11Domestic carbon pricing 4.5
  • Linkage and exchangen=7Border carbon adjustment 4.3
  • Industry and policy referencen=5Border carbon adjustment 3.8
Q67 Ranked top 3 n = 36

Rank the three parameters that contribute most to uncertainty in the 2035 cost of hydrogen-based steel.

Renewable electricity price and firming: 71% · 1st 58% Renewable electricity price and firming 71% · 1st 58% Green premium buyers will pay: 29% · 1st 11% Green premium buyers will pay 29% · 1st 11% Carbon price and CBAM: 26% · 1st 14% Carbon price and CBAM 26% · 1st 14% Electrolyser capex and efficiency: 21% · 1st 0% Electrolyser capex and efficiency 21% · 1st 0% Natural gas price of the competing route: 16% · 1st 6% Natural gas price of the competing route 16% · 1st 6% Cost of capital: 12% · 1st 6% Cost of capital 12% · 1st 6% Plant utilisation and flexibility costs: 12% · 1st 6% Plant utilisation and flexibility costs 12% · 1st 6% DR-grade ore premium: 10% · 1st 0% DR-grade ore premium 10% · 1st 0% DRI and EAF plant capex: 3% · 1st 0% DRI and EAF plant capex 3% · 1st 0%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 36 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Renewable electricity price and firming71%58%81%
Green premium buyers will pay29%11%44%
Carbon price and CBAM26%14%36%
Electrolyser capex and efficiency21%0%39%
Natural gas price of the competing route16%6%31%
Cost of capital12%6%19%
Plant utilisation and flexibility costs12%6%22%
DR-grade ore premium10%0%22%
DRI and EAF plant capex3%0%6%

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=5Renewable electricity price and firming (Borda 87%)
  • Subtask Cn=13Renewable electricity price and firming (Borda 72%)
  • Linkage and exchangen=7Renewable electricity price and firming (Borda 67%)
  • Industry and policy referencen=6Renewable electricity price and firming (Borda 56%)
Q68 Year percentiles n = 27

In which year will hydrogen-based DRI-EAF crude steel made in Europe become cheaper than unabated BF-BOF steel in Europe, including carbon pricing and border adjustment but excluding subsidies?

Give the year by which you assign a 10%, 50% and 90% chance. Fill the 10% and 90% years first.

10% chance by

2033

IQR 2031–2035

50% chance by

2040

IQR 2037–2043

90% chance by

2048

IQR 2044–2050

0%25%50%75%100%20272031203520392043204720502027: 1% pooled probability2028: 4% pooled probability2029: 6% pooled probability2030: 8% pooled probability2031: 11% pooled probability2032: 13% pooled probability2033: 17% pooled probability2034: 21% pooled probability2035: 26% pooled probability2036: 31% pooled probability2037: 36% pooled probability2038: 42% pooled probability2039: 47% pooled probability2040: 53% pooled probability2041: 57% pooled probability2042: 62% pooled probability2043: 66% pooled probability2044: 71% pooled probability2045: 75% pooled probability2046: 78% pooled probability2047: 81% pooled probability2048: 84% pooled probability2049: 87% pooled probability2050: 90% pooled probability

Pooled cumulative probability: average of each expert's piecewise-linear curve through their 10%, 50% and 90% years.

“After 2050” is counted as 2055 for the statistics
PercentilenMedian yearIQRRange“Never”
10% chance by2720332031 to 20352027 to 20420
50% chance by2720402037 to 20432028 to 20550
90% chance by2720482044 to 20502030 to 20550

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=1150% chance by 2040
  • Linkage and exchangen=650% chance by 2038
Q69 Low, best and high estimate n = 30

What will global crude steel demand be in 2050, all routes and regions?

Published scenarios differ by more than 200 Mt: IEA Net Zero around 1,960 Mt after material-efficiency savings, Agora Industry around 2,175 Mt; 2025 demand is about 1,750 Mt.

Median low

1800.0

Median best

2025.0

IQR 1905.0–2200.0

Median high

2315.0

120015201840216024802800Interquartile range of best estimates: 1905.0–2200.0Median best estimate 2025.0Expert: low 1200, best 1200, high 1200Expert: low 1200, best 1250, high 1500Expert: low 1500, best 1800, high 2200Expert: low 1700, best 1850, high 1900Expert: low 1240, best 1880, high 2110Expert: low 1600, best 1880, high 2200Expert: low 1700, best 1890, high 2200Expert: low 1800, best 1900, high 2100Expert: low 1910, best 1920, high 2010Expert: low 1500, best 2000, high 2200Expert: low 1600, best 2000, high 2400Expert: low 1700, best 2000, high 2200Expert: low 1750, best 2000, high 2300Expert: low 1800, best 2000, high 2200Expert: low 1800, best 2000, high 2500Expert: low 1800, best 2050, high 2300Expert: low 1750, best 2100, high 2500Expert: low 1800, best 2100, high 2500Expert: low 1920, best 2100, high 2240Expert: low 1900, best 2140, high 2400Expert: low 1600, best 2200, high 2330Expert: low 1800, best 2200, high 2510Expert: low 1900, best 2200, high 2400Expert: low 2000, best 2200, high 2400Expert: low 2000, best 2200, high 2350Expert: low 2000, best 2200, high 2500Expert: low 2010, best 2310, high 2680Expert: low 1900, best 2400, high 2800Expert: low 2000, best 2500, high 2800Expert: low 2400, best 2600, high 2800

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: Mt per year of crude steel

Unit: Mt per year of crude steel
EstimatenMedianIQRRangeMean
Lowest plausible3018001625 to 19081200 to 24001759.3
Best estimate3020251905 to 22001200 to 26002035.7
Highest plausible3023152200 to 25001200 to 28002291.0

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=5best 2000 Mt per year of crude steel
  • Subtask Cn=12best 2050 Mt per year of crude steel
  • Linkage and exchangen=7best 2000 Mt per year of crude steel
Q70 Choice grid n = 24

What will the average delivered price of firmed, near-continuous renewable electricity be for a large flexible industrial user such as a green iron plant in 2030, in each region?

Europe, best-in-class site Europe, best-in-class site · Below USD 30/MWh: 10% (n=21)Europe, best-in-class site · 30 to 50: 38% (n=21)Europe, best-in-class site · 50 to 70: 19% (n=21)Europe, best-in-class site · 70 to 100: 24% (n=21)Europe, best-in-class site · Above USD 100/MWh: 10% (n=21) Mean score (n=21)60.2 Australia, best-in-class site Australia, best-in-class site · Below USD 30/MWh: 35% (n=20)Australia, best-in-class site · 30 to 50: 15% (n=20)Australia, best-in-class site · 50 to 70: 30% (n=20)Australia, best-in-class site · 70 to 100: 15% (n=20)Australia, best-in-class site · Above USD 100/MWh: 5% (n=20) Mean score (n=20)51.3 Middle East and North Africa Middle East and North Africa · Below USD 30/MWh: 35% (n=20)Middle East and North Africa · 30 to 50: 45% (n=20)Middle East and North Africa · 50 to 70: 15% (n=20)Middle East and North Africa · Above USD 100/MWh: 5% (n=20) Mean score (n=20)41.5 United States United States · Below USD 30/MWh: 16% (n=19)United States · 30 to 50: 47% (n=19)United States · 50 to 70: 26% (n=19)United States · 70 to 100: 5% (n=19)United States · Above USD 100/MWh: 5% (n=19) Mean score (n=19)49.2 China China · Below USD 30/MWh: 43% (n=21)China · 30 to 50: 38% (n=21)China · 50 to 70: 5% (n=21)China · 70 to 100: 10% (n=21)China · Above USD 100/MWh: 5% (n=21) Mean score (n=21)42.4 India India · Below USD 30/MWh: 32% (n=19)India · 30 to 50: 21% (n=19)India · 50 to 70: 37% (n=19)India · 70 to 100: 5% (n=19)India · Above USD 100/MWh: 5% (n=19) Mean score (n=19)48.9
Below USD 30/MWh30 to 5050 to 7070 to 100Above USD 100/MWhNumber at right: mean score
Share of the experts answering each row
RownBelow USD 30/MWh30 to 5050 to 7070 to 100Above USD 100/MWh
Europe, best-in-class site2110%38%19%24%10%
Australia, best-in-class site2035%15%30%15%5%
Middle East and North Africa2035%45%15%0%5%
United States1916%47%26%5%5%
China2143%38%5%10%5%
India1932%21%37%5%5%

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=9Europe, best-in-class site: 50 to 70 38%
  • Linkage and exchangen=5Europe, best-in-class site: 30 to 50 75%
Q71 Low, best and high estimate n = 24

What share of global hydrogen consumption will iron and steel production account for in 2035, all routes, including hydrogen used via ammonia?

Today iron and steel take roughly 5% of the world's hydrogen use of about 100 Mt per year, almost all of it in natural-gas DRI. Published studies cluster around 2050 endpoints; no comparable 2035 share exists, so the network's estimate fills a real gap.

Median low

6.0

Median best

11.0

IQR 10.0–18.5

Median high

20.0

01836547290Interquartile range of best estimates: 10.0–18.5Median best estimate 11.0Expert: low 0, best 2, high 5Expert: low 3, best 5, high 15Expert: low 4, best 7, high 13Expert: low 5, best 8, high 10Expert: low 3, best 8.5, high 14.5Expert: low 5, best 10, high 25.5Expert: low 5, best 10, high 15Expert: low 5, best 10, high 15Expert: low 5, best 10, high 30Expert: low 6, best 10, high 20Expert: low 7, best 10, high 15Expert: low 6, best 11, high 18Expert: low 6.5, best 11, high 13Expert: low 8, best 13.5, high 23.5Expert: low 5, best 15, high 25Expert: low 10, best 15, high 15Expert: low 12, best 17, high 20Expert: low 9, best 18, high 21.5Expert: low 5, best 20, high 40Expert: low 20.5, best 25, high 40Expert: low 30, best 35, high 50Expert: low 22, best 40, high 55Expert: low 20, best 50, high 90Expert: low 43.5, best 54.5, high 68

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: % of global hydrogen consumption

Unit: % of global hydrogen consumption
EstimatenMedianIQRRangeMean
Lowest plausible2465 to 110 to 4410.2
Best estimate241110 to 192 to 5517.3
Highest plausible242015 to 335 to 9027.4

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=8best 14% of global hydrogen consumption
  • Linkage and exchangen=6best 11% of global hydrogen consumption
Q72 Ranked top 3 n = 36

Which investments are least-regret for a steelmaker deciding today, meaning they hold value across most hydrogen-price and policy scenarios out to 2040? Rank your top three.

Scrap-based EAF capacity and scrap sourcing: 75% · 1st 61% Scrap-based EAF capacity and scrap sourcing 75% · 1st 61% Grid connection and long-term renewable power contracts: 26% · 1st 6% Grid connection and long-term renewable power… 26% · 1st 6% Hydrogen-ready DRI started on natural gas: 25% · 1st 8% Hydrogen-ready DRI started on natural gas 25% · 1st 8% Long-term green-steel offtake contracts with buyers: 23% · 1st 6% Long-term green-steel offtake contracts with … 23% · 1st 6% Electric smelter (ESF/OSBF) capability for lower-grade ore: 13% · 1st 6% Electric smelter (ESF/OSBF) capability for lo… 13% · 1st 6% Keeping blast furnaces running with CCUS or hydrogen-injection readiness: 13% · 1st 6% Keeping blast furnaces running with CCUS or h… 13% · 1st 6% Securing DR-grade ore or beneficiation capacity: 8% · 1st 3% Securing DR-grade ore or beneficiation capaci… 8% · 1st 3% Import agreements for green HBI or merchant iron: 8% · 1st 3% Import agreements for green HBI or merchant i… 8% · 1st 3% Deferring major capital decisions until after 2030: 8% · 1st 3% Deferring major capital decisions until after… 8% · 1st 3%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 36 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Scrap-based EAF capacity and scrap sourcing75%61%89%
Grid connection and long-term renewable power contracts26%6%44%
Hydrogen-ready DRI started on natural gas25%8%39%
Long-term green-steel offtake contracts with buyers23%6%39%
Electric smelter (ESF/OSBF) capability for lower-grade ore13%6%22%
Keeping blast furnaces running with CCUS or hydrogen-injection readiness13%6%19%
Securing DR-grade ore or beneficiation capacity8%3%17%
Import agreements for green HBI or merchant iron8%3%14%
Deferring major capital decisions until after 20308%3%17%

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=5Scrap-based EAF capacity and scrap sourcing (Borda 67%)
  • Subtask Cn=13Scrap-based EAF capacity and scrap sourcing (Borda 85%)
  • Linkage and exchangen=7Scrap-based EAF capacity and scrap sourcing (Borda 81%)
  • Industry and policy referencen=6Scrap-based EAF capacity and scrap sourcing (Borda 72%)
Q73 Ranked top 3 n = 33

An existing integrated BF-BOF site in a high-cost jurisdiction faces a blast-furnace reline decision before 2032. Rank the three least-regret choices, least regret first.

Depends entirely on the site; no general answer: 63% · 1st 58% Depends entirely on the site; no general answ… 63% · 1st 58% Replace with EAF fed by scrap and imported HBI: 52% · 1st 15% Replace with EAF fed by scrap and imported HBI 52% · 1st 15% Replace with hydrogen-ready natural-gas DRI and EAF: 24% · 1st 6% Replace with hydrogen-ready natural-gas DRI a… 24% · 1st 6% Retire the capacity and source slab or iron elsewhere: 23% · 1st 6% Retire the capacity and source slab or iron e… 23% · 1st 6% Reline unabated and decide later: 20% · 1st 6% Reline unabated and decide later 20% · 1st 6% Reline with hydrogen-injection or CCUS readiness: 18% · 1st 9% Reline with hydrogen-injection or CCUS readin… 18% · 1st 9%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 33 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
Depends entirely on the site; no general answer63%58%70%
Replace with EAF fed by scrap and imported HBI52%15%82%
Replace with hydrogen-ready natural-gas DRI and EAF24%6%48%
Retire the capacity and source slab or iron elsewhere23%6%39%
Reline unabated and decide later20%6%33%
Reline with hydrogen-injection or CCUS readiness18%9%27%

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=11Depends entirely on the site; no general answer (Borda 76%)
  • Linkage and exchangen=7Replace with EAF fed by scrap and imported HBI (Borda 76%)
  • Industry and policy referencen=6Depends entirely on the site; no general answer (Borda 83%)
Q74 Choice grid n = 28

How robust is each pathway if renewable hydrogen still costs more than USD 4 per kilogram in 2035?

Hydrogen DRI at the existing steel site Hydrogen DRI at the existing steel site · Robust: still the right choice: 7% (n=28)Hydrogen DRI at the existing steel site · Tolerable: viable with support: 46% (n=28)Hydrogen DRI at the existing steel site · At risk of stranding: 46% (n=28) Mean score (n=28)0.6 Importing green HBI from ore-and-renewa… Importing green HBI from ore-and-renewables hubs · Robust: still the right choice: 26% (n=27)Importing green HBI from ore-and-renewables hubs · Tolerable: viable with support: 63% (n=27)Importing green HBI from ore-and-renewables hubs · At risk of stranding: 11% (n=27) Mean score (n=27)1.1 Natural-gas DRI with CCS Natural-gas DRI with CCS · Robust: still the right choice: 35% (n=26)Natural-gas DRI with CCS · Tolerable: viable with support: 38% (n=26)Natural-gas DRI with CCS · At risk of stranding: 27% (n=26) Mean score (n=26)1.1 Blast furnace with CCUS Blast furnace with CCUS · Robust: still the right choice: 21% (n=24)Blast furnace with CCUS · Tolerable: viable with support: 21% (n=24)Blast furnace with CCUS · At risk of stranding: 58% (n=24) Mean score (n=24)0.6 Scrap-based EAF expansion Scrap-based EAF expansion · Robust: still the right choice: 85% (n=27)Scrap-based EAF expansion · Tolerable: viable with support: 15% (n=27) Mean score (n=27)1.9 Electric smelter on lower-grade ore wit… Electric smelter on lower-grade ore with hydrogen DRI · Robust: still the right choice: 16% (n=25)Electric smelter on lower-grade ore with hydrogen DRI · Tolerable: viable with support: 48% (n=25)Electric smelter on lower-grade ore with hydrogen DRI · At risk of stranding: 36% (n=25) Mean score (n=25)0.8
Robust: still the right choiceTolerable: viable with supportAt risk of strandingNumber at right: mean score
Share of the experts answering each row
RownRobust: still the right choiceTolerable: viable with supportAt risk of stranding
Hydrogen DRI at the existing steel site287%46%46%
Importing green HBI from ore-and-renewables hubs2726%63%11%
Natural-gas DRI with CCS2635%38%27%
Blast furnace with CCUS2421%21%58%
Scrap-based EAF expansion2785%15%0%
Electric smelter on lower-grade ore with hydrogen DRI2516%48%36%

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=9Hydrogen DRI at the existing steel site: At risk of stranding 67%
  • Linkage and exchangen=6Hydrogen DRI at the existing steel site: Tolerable: viable with support 50%
Q75 Ranked top 3 n = 36

Which policy measures are least-regret, meaning they help the transition whichever technology route wins? Rank your top three.

A rising, predictable carbon price: 65% · 1st 56% A rising, predictable carbon price 65% · 1st 56% Lead markets: public procurement and quotas for near-zero steel: 29% · 1st 8% Lead markets: public procurement and quotas f… 29% · 1st 8% Grid build-out and faster connection for industrial loads: 26% · 1st 8% Grid build-out and faster connection for indu… 26% · 1st 8% Border carbon adjustment on imported steel and iron: 24% · 1st 6% Border carbon adjustment on imported steel an… 24% · 1st 6% Hydrogen production and transport infrastructure: 16% · 1st 11% Hydrogen production and transport infrastruct… 16% · 1st 11% Common emissions accounting and product standards: 15% · 1st 0% Common emissions accounting and product stand… 15% · 1st 0% Scrap retention and quality policies: 12% · 1st 3% Scrap retention and quality policies 12% · 1st 3% Carbon contracts for difference: 11% · 1st 6% Carbon contracts for difference 11% · 1st 6% Research and demonstration funding: 3% · 1st 3% Research and demonstration funding 3% · 1st 3%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 36 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
A rising, predictable carbon price65%56%72%
Lead markets: public procurement and quotas for near-zero steel29%8%50%
Grid build-out and faster connection for industrial loads26%8%47%
Border carbon adjustment on imported steel and iron24%6%39%
Hydrogen production and transport infrastructure16%11%19%
Common emissions accounting and product standards15%0%28%
Scrap retention and quality policies12%3%25%
Carbon contracts for difference11%6%17%
Research and demonstration funding3%3%3%

By Task 52 affiliation (groups of at least 5)

  • Subtask Bn=5A rising, predictable carbon price (Borda 40%)
  • Subtask Cn=13A rising, predictable carbon price (Borda 77%)
  • Linkage and exchangen=7A rising, predictable carbon price (Borda 71%)
  • Industry and policy referencen=6A rising, predictable carbon price (Borda 56%)
Regional pathways 42 experts · 7 questions
Q76 Allocation of 100 points n = 28

Allocate China's domestic crude steel production in 2035 across production routes (percent, summing to 100).

China produced roughly 1,000 Mt in 2025, about 90% via BF-BOF.

0%20%40%60%80%100% BF-BOF, unabated: median 62.0%, IQR 50.0–70.0, range 17.0–81.0 (n=28) BF-BOF, unabated 62 BF-BOF with CCUS: median 5.0%, IQR 3.0–8.8, range 1.0–15.0 (n=22) BF-BOF with CCUS 5 Natural-gas DRI-EAF: median 1.0%, IQR 0.0–5.0, range 0.0–27.0 (n=18) Natural-gas DRI-EAF 1 Hydrogen-based DRI-EAF: median 9.0%, IQR 3.0–10.0, range 1.0–70.0 (n=26) Hydrogen-based DRI-EAF 9 Scrap-based EAF: median 22.0%, IQR 15.8–28.5, range 8.0–50.0 (n=28) Scrap-based EAF 22 Other novel routes: median 1.0%, IQR 0.0–2.8, range 0.0–10.0 (n=16) Other novel routes 1

Box: interquartile range · thick line: median · whiskers: min–max · number: median

Points out of 100 · reference values are the baseline shown to experts, where one was given
OptionnMedianIQRRangeReference
BF-BOF, unabated2862.050.0 to 70.017 to 81–
BF-BOF with CCUS225.03.0 to 8.81 to 15–
Natural-gas DRI-EAF181.00.0 to 5.00 to 27–
Hydrogen-based DRI-EAF269.03.0 to 10.01 to 70–
Scrap-based EAF2822.015.8 to 28.58 to 50–
Other novel routes161.00.0 to 2.80 to 10–

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5BF-BOF, unabated 80
  • Subtask Cn=7BF-BOF, unabated 68
  • Linkage and exchangen=8BF-BOF, unabated 58
  • Industry and policy referencen=5BF-BOF, unabated 60
Q77 Single choice n = 31

By 2035, will China be a net exporter or a net importer of hydrogen-based DRI and HBI, by trade volume?

Significant net exporter, more than 5 Mt per year: 3% (1) Significant net exporter, more than 5 Mt per … 3% (1) Modest net exporter: 35% (11) Modest net exporter 35% (11) Roughly balanced: 16% (5) Roughly balanced 16% (5) Modest net importer: 26% (8) Modest net importer 26% (8) Significant net importer, more than 5 Mt per year: 3% (1) Significant net importer, more than 5 Mt per … 3% (1) China's green iron trade stays negligible either way: 16% (5) China's green iron trade stays negligible eit… 16% (5)

Mean score 0.1, median 0.0

n = 31 · mean score 0.10, median 0.0
OptionExpertsShare
Significant net exporter, more than 5 Mt per year13%
Modest net exporter1135%
Roughly balanced516%
Modest net importer826%
Significant net importer, more than 5 Mt per year13%
China's green iron trade stays negligible either way516%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5Modest net importer 40%
  • Subtask Cn=7Modest net importer 43%
  • Linkage and exchangen=8Modest net exporter 38%
  • Industry and policy referencen=6Modest net exporter 50%
Q78 Single choice n = 27

Which will be the dominant direct-reduction pathway in India's steel production, by tonnage, in 2035?

Coal-based rotary-kiln DRI remains dominant: 52% (14) Coal-based rotary-kiln DRI remains dominant 52% (14) Natural-gas shaft-furnace DRI grows to dominate: 7% (2) Natural-gas shaft-furnace DRI grows to domina… 7% (2) Hydrogen-based DRI reaches commercial scale and dominates new capacity: 11% (3) Hydrogen-based DRI reaches commercial scale a… 11% (3) BF-BOF expansion outpaces all DRI routes: 30% (8) BF-BOF expansion outpaces all DRI routes 30% (8) No single route dominates: 0% (0) No single route dominates 0% (0)
n = 27
OptionExpertsShare
Coal-based rotary-kiln DRI remains dominant1452%
Natural-gas shaft-furnace DRI grows to dominate27%
Hydrogen-based DRI reaches commercial scale and dominates new capacity311%
BF-BOF expansion outpaces all DRI routes830%
No single route dominates00%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5BF-BOF expansion outpaces all DRI routes 60%
  • Subtask Cn=6Coal-based rotary-kiln DRI remains dominant 67%
  • Linkage and exchangen=6Coal-based rotary-kiln DRI remains dominant 67%
  • Industry and policy referencen=5BF-BOF expansion outpaces all DRI routes 60%
Q79 Single choice n = 24

Which route will account for most new primary ironmaking capacity built in the United States between 2025 and 2035?

Natural-gas DRI-EAF with CCS: 21% (5) Natural-gas DRI-EAF with CCS 21% (5) Natural-gas DRI-EAF without CCS: 50% (12) Natural-gas DRI-EAF without CCS 50% (12) Hydrogen-based DRI-EAF: 0% (0) Hydrogen-based DRI-EAF 0% (0) BF-BOF, new or relined, unabated: 0% (0) BF-BOF, new or relined, unabated 0% (0) Scrap-EAF absorbs nearly all growth; little new primary capacity: 29% (7) Scrap-EAF absorbs nearly all growth; little n… 29% (7)
n = 24
OptionExpertsShare
Natural-gas DRI-EAF with CCS521%
Natural-gas DRI-EAF without CCS1250%
Hydrogen-based DRI-EAF00%
BF-BOF, new or relined, unabated00%
Scrap-EAF absorbs nearly all growth; little new primary capacity729%

By Task 52 affiliation (groups of at least 5)

  • Subtask Cn=6Natural-gas DRI-EAF without CCS 50%
  • Linkage and exchangen=5Natural-gas DRI-EAF without CCS 60%
  • Industry and policy referencen=5Natural-gas DRI-EAF without CCS 60%
Q80 Low, best and high estimate n = 25

Of the EU blast-furnace capacity that needs a reinvestment decision by 2030, about three quarters of the fleet, what share will be permanently closed by 2035 with no low-emission replacement built on the same site?

Industry tracking puts the share of EU blast-furnace capacity due for a major reinvestment (relining) decision by 2030 at roughly 70% to 75%: Agora Energiewende's Global Steel Transformation Tracker, as cited by the Climate Bonds Initiative (2022), gives 74%, about 74 Mt; Bellona Europa (2024) says 70% of Europe's steelmaking facilities need major investment before 2030. Earlier Agora analysis (2021) put it at about half. Answer as a share of that at-risk capacity.

Median low

22.0

Median best

49.0

IQR 25.0–55.0

Median high

71.0

020406080100Interquartile range of best estimates: 25.0–55.0Median best estimate 49.0Expert: low 5, best 8, high 18Expert: low 0, best 10, high 15Expert: low 5, best 20, high 40Expert: low 15, best 20, high 50Expert: low 20, best 20, high 26Expert: low 20, best 25, high 30Expert: low 20, best 25, high 50Expert: low 20, best 35, high 50Expert: low 20, best 35, high 60Expert: low 0, best 40, high 100Expert: low 20, best 40, high 60Expert: low 33, best 47, high 61Expert: low 25, best 49, high 100Expert: low 20, best 50, high 90Expert: low 25, best 50, high 75Expert: low 25, best 50, high 75Expert: low 25, best 50, high 60Expert: low 25, best 50, high 100Expert: low 25, best 55, high 75Expert: low 22, best 56, high 71Expert: low 25, best 60, high 75Expert: low 50, best 60, high 80Expert: low 50, best 65, high 75Expert: low 50, best 70, high 80Expert: low 70, best 75, high 75

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: % of at-risk EU blast-furnace capacity

Unit: % of at-risk EU blast-furnace capacity
EstimatenMedianIQRRangeMean
Lowest plausible252220 to 250 to 7024.6
Best estimate254925 to 558 to 7542.6
Highest plausible257150 to 7515 to 10063.6

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5best 40% of at-risk EU blast-furnace capacity
  • Subtask Cn=7best 50% of at-risk EU blast-furnace capacity
  • Linkage and exchangen=7best 50% of at-risk EU blast-furnace capacity
Q81 Probability, 0 to 100% n = 33

By 2035, what share of each region's primary iron units for near-zero steel will be imported as HBI or DRI rather than reduced domestically? Use the slider for each region.

One value per region, from 0% (all reduced domestically) to 100% (all imported). Agora Industry and the Renewable Energy Institute estimate imported green HBI could be about 15% cheaper than domestic hydrogen DRI in Japan.

0%20%40%60%80%100%Japan: median 45%, IQR 10–63 (n=24) Japan 45%South Korea: median 35%, IQR 5–65 (n=23) South Korea 35%European Union: median 25%, IQR 20–45 (n=27) European Union 25%China: median 5%, IQR 0–14 (n=26) China 5%India: median 5%, IQR 0–11 (n=24) India 5%

Dot: median · bar: interquartile range

Probability sliders, 0 to 100% · rows sorted by mean
RownMedian %IQR %Range %
Japan244510 to 630 to 100
South Korea23355 to 650 to 100
European Union272520 to 450 to 90
China2650 to 140 to 100
India2450 to 110 to 75

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5Japan 90%
  • Subtask Cn=8South Korea 40%
  • Linkage and exchangen=9European Union 25%
  • Industry and policy referencen=7South Korea 40%
Q82 Probability, 0 to 100% n = 32

And by 2050, what share of each region's primary iron units for near-zero steel will be imported as HBI or DRI rather than reduced domestically? Use the slider for each region.

One value per region, from 0% (all reduced domestically) to 100% (all imported). Judge the mature trade pattern once low-cost renewable regions and ore exporters have built out reduction capacity.

0%20%40%60%80%100%Japan: median 65%, IQR 50–80 (n=25) Japan 65%South Korea: median 60%, IQR 33–70 (n=23) South Korea 60%European Union: median 50%, IQR 45–60 (n=26) European Union 50%China: median 10%, IQR 5–20 (n=26) China 10%India: median 10%, IQR 5–20 (n=25) India 10%

Dot: median · bar: interquartile range

Probability sliders, 0 to 100% · rows sorted by mean
RownMedian %IQR %Range %
Japan256550 to 8010 to 100
South Korea236033 to 700 to 100
European Union265045 to 6010 to 95
China26105 to 200 to 70
India25105 to 200 to 75

By Task 52 affiliation (groups of at least 5)

  • Subtask An=5Japan 88%
  • Subtask Cn=8South Korea 83%
  • Linkage and exchangen=8Japan 55%
  • Industry and policy referencen=7European Union 60%
Markets, standards and policy 81 experts · 11 questions
Q83 Single choice n = 66

What price premium per tonne will buyers sustainably pay for near-zero emission steel in 2030, averaged across segments?

No premium: 9% (6) No premium 9% (6) Up to USD 50: 32% (21) Up to USD 50 32% (21) USD 50 to 100: 35% (23) USD 50 to 100 35% (23) USD 100 to 200: 18% (12) USD 100 to 200 18% (12) More than USD 200: 6% (4) More than USD 200 6% (4)

Mean score 76.5, median 75.0

n = 66 · mean score 76.52, median 75.0
OptionExpertsShare
No premium69%
Up to USD 502132%
USD 50 to 1002335%
USD 100 to 2001218%
More than USD 20046%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=17USD 50 to 100 41%
  • Subtask Bn=12USD 50 to 100 42%
  • Subtask Cn=9Up to USD 50 56%
  • Linkage and exchangen=10USD 50 to 100 60%
  • Industry and policy referencen=14Up to USD 50 43%
Q84 Choice grid n = 68

How far will each end-use sector have moved to near-zero emission steel by 2035?

Automotive Automotive · Pilots only: 3% (n=68)Automotive · Slow progress: 41% (n=68)Automotive · Rapid progress: 40% (n=68)Automotive · Largely transitioned: 16% (n=68) Mean score (n=68)2.7 Construction and infrastructure Construction and infrastructure · No change: 12% (n=67)Construction and infrastructure · Pilots only: 22% (n=67)Construction and infrastructure · Slow progress: 60% (n=67)Construction and infrastructure · Rapid progress: 6% (n=67) Mean score (n=67)1.6 Wind, solar and grid equipment Wind, solar and grid equipment · No change: 6% (n=66)Wind, solar and grid equipment · Pilots only: 15% (n=66)Wind, solar and grid equipment · Slow progress: 33% (n=66)Wind, solar and grid equipment · Rapid progress: 36% (n=66)Wind, solar and grid equipment · Largely transitioned: 9% (n=66) Mean score (n=66)2.3 Appliances and consumer goods Appliances and consumer goods · No change: 15% (n=66)Appliances and consumer goods · Pilots only: 20% (n=66)Appliances and consumer goods · Slow progress: 44% (n=66)Appliances and consumer goods · Rapid progress: 18% (n=66)Appliances and consumer goods · Largely transitioned: 3% (n=66) Mean score (n=66)1.7 Shipbuilding Shipbuilding · No change: 20% (n=65)Shipbuilding · Pilots only: 29% (n=65)Shipbuilding · Slow progress: 37% (n=65)Shipbuilding · Rapid progress: 14% (n=65) Mean score (n=65)1.4 Machinery and heavy equipment Machinery and heavy equipment · No change: 22% (n=64)Machinery and heavy equipment · Pilots only: 27% (n=64)Machinery and heavy equipment · Slow progress: 44% (n=64)Machinery and heavy equipment · Rapid progress: 8% (n=64) Mean score (n=64)1.4
No changePilots onlySlow progressRapid progressLargely transitionedNumber at right: mean score
Share of the experts answering each row
RownNo changePilots onlySlow progressRapid progressLargely transitioned
Automotive680%3%41%40%16%
Construction and infrastructure6712%22%60%6%0%
Wind, solar and grid equipment666%15%33%36%9%
Appliances and consumer goods6615%20%44%18%3%
Shipbuilding6520%29%37%14%0%
Machinery and heavy equipment6422%27%44%8%0%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=16Automotive: Rapid progress 50%
  • Subtask Bn=13Automotive: Rapid progress 62%
  • Subtask Cn=9Automotive: Slow progress 78%
  • Linkage and exchangen=10Automotive: Slow progress 40%
  • Industry and policy referencen=16Automotive: Slow progress 38%
Q85 Rating, 1 (Not limiting) to 5 (Severely limiting) n = 63

How much does each knowledge gap limit decisions by industry and governments today?

Carbon accounting and certification rul… Carbon accounting and certification rules · 2: 7% (n=60)Carbon accounting and certification rules · 3: 20% (n=60)Carbon accounting and certification rules · 4: 48% (n=60)Carbon accounting and certification rules · 5: 25% (n=60) Mean score (n=60)3.9 Comparison of national policy packages Comparison of national policy packages · 1: 2% (n=57)Comparison of national policy packages · 2: 11% (n=57)Comparison of national policy packages · 3: 33% (n=57)Comparison of national policy packages · 4: 37% (n=57)Comparison of national policy packages · 5: 18% (n=57) Mean score (n=57)3.6 Power-system integration assessment Power-system integration assessment · 1: 5% (n=59)Power-system integration assessment · 2: 12% (n=59)Power-system integration assessment · 3: 29% (n=59)Power-system integration assessment · 4: 32% (n=59)Power-system integration assessment · 5: 22% (n=59) Mean score (n=59)3.5 Plant flexibility and hydrogen bufferin… Plant flexibility and hydrogen buffering requirements · 1: 3% (n=61)Plant flexibility and hydrogen buffering requirements · 2: 10% (n=61)Plant flexibility and hydrogen buffering requirements · 3: 33% (n=61)Plant flexibility and hydrogen buffering requirements · 4: 39% (n=61)Plant flexibility and hydrogen buffering requirements · 5: 15% (n=61) Mean score (n=61)3.5 Operating data from demonstration plants Operating data from demonstration plants · 1: 3% (n=63)Operating data from demonstration plants · 2: 21% (n=63)Operating data from demonstration plants · 3: 19% (n=63)Operating data from demonstration plants · 4: 38% (n=63)Operating data from demonstration plants · 5: 19% (n=63) Mean score (n=63)3.5 Trade and relocation scenarios Trade and relocation scenarios · 1: 2% (n=58)Trade and relocation scenarios · 2: 17% (n=58)Trade and relocation scenarios · 3: 33% (n=58)Trade and relocation scenarios · 4: 29% (n=58)Trade and relocation scenarios · 5: 19% (n=58) Mean score (n=58)3.5 Comparative supply-chain cost model Comparative supply-chain cost model · 2: 21% (n=56)Comparative supply-chain cost model · 3: 25% (n=56)Comparative supply-chain cost model · 4: 39% (n=56)Comparative supply-chain cost model · 5: 14% (n=56) Mean score (n=56)3.5 Processing routes for low-grade ores Processing routes for low-grade ores · 1: 5% (n=62)Processing routes for low-grade ores · 2: 15% (n=62)Processing routes for low-grade ores · 3: 31% (n=62)Processing routes for low-grade ores · 4: 31% (n=62)Processing routes for low-grade ores · 5: 19% (n=62) Mean score (n=62)3.5 Harmonised techno-economic dataset acro… Harmonised techno-economic dataset across routes · 1: 5% (n=61)Harmonised techno-economic dataset across routes · 2: 25% (n=61)Harmonised techno-economic dataset across routes · 3: 34% (n=61)Harmonised techno-economic dataset across routes · 4: 21% (n=61)Harmonised techno-economic dataset across routes · 5: 15% (n=61) Mean score (n=61)3.2 Workforce and skills requirements Workforce and skills requirements · 1: 5% (n=58)Workforce and skills requirements · 2: 29% (n=58)Workforce and skills requirements · 3: 45% (n=58)Workforce and skills requirements · 4: 17% (n=58)Workforce and skills requirements · 5: 3% (n=58) Mean score (n=58)2.8
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Carbon accounting and certification rules603.924.03.0 to 4.3
Comparison of national policy packages573.584.03.0 to 4.0
Power-system integration assessment593.544.03.0 to 4.0
Plant flexibility and hydrogen buffering requirements613.524.03.0 to 4.0
Operating data from demonstration plants633.494.03.0 to 4.0
Trade and relocation scenarios583.473.03.0 to 4.0
Comparative supply-chain cost model563.464.03.0 to 4.0
Processing routes for low-grade ores623.453.53.0 to 4.0
Harmonised techno-economic dataset across routes613.163.02.0 to 4.0
Workforce and skills requirements582.843.02.0 to 3.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=19Carbon accounting and certification rules 3.8
  • Subtask Bn=12Carbon accounting and certification rules 4.4
  • Subtask Cn=8Carbon accounting and certification rules 3.9
  • Linkage and exchangen=10Carbon accounting and certification rules 3.8
  • Industry and policy referencen=11Comparison of national policy packages 3.8
Q86 Rating, 1 (Not important) to 5 (Essential) n = 67

How important is each safeguard for emissions-accounting schemes covering near-zero and low-emission steel?

1 = not important, 5 = essential.

Prevention of double counting of emissi… Prevention of double counting of emission reductions · 2: 6% (n=66)Prevention of double counting of emission reductions · 3: 8% (n=66)Prevention of double counting of emission reductions · 4: 24% (n=66)Prevention of double counting of emission reductions · 5: 62% (n=66) Mean score (n=66)4.4 Transparent eligibility and boundary ru… Transparent eligibility and boundary rules · 2: 1% (n=67)Transparent eligibility and boundary rules · 3: 12% (n=67)Transparent eligibility and boundary rules · 4: 30% (n=67)Transparent eligibility and boundary rules · 5: 57% (n=67) Mean score (n=67)4.4 Mutual recognition between schemes acro… Mutual recognition between schemes across borders · 2: 3% (n=66)Mutual recognition between schemes across borders · 3: 9% (n=66)Mutual recognition between schemes across borders · 4: 39% (n=66)Mutual recognition between schemes across borders · 5: 48% (n=66) Mean score (n=66)4.3 Public disclosure of the allocation met… Public disclosure of the allocation method used · 2: 1% (n=67)Public disclosure of the allocation method used · 3: 15% (n=67)Public disclosure of the allocation method used · 4: 37% (n=67)Public disclosure of the allocation method used · 5: 46% (n=67) Mean score (n=67)4.3 Physical traceability or chain of custo… Physical traceability or chain of custody for the product · 2: 2% (n=66)Physical traceability or chain of custody for the product · 3: 14% (n=66)Physical traceability or chain of custody for the product · 4: 44% (n=66)Physical traceability or chain of custody for the product · 5: 41% (n=66) Mean score (n=66)4.2 Independent third-party verification Independent third-party verification · 3: 22% (n=67)Independent third-party verification · 4: 34% (n=67)Independent third-party verification · 5: 43% (n=67) Mean score (n=67)4.2
12345Number at right: mean score
Scale 1 to 5 · rows sorted by mean
RownMeanMedianIQR
Prevention of double counting of emission reductions664.425.04.0 to 5.0
Transparent eligibility and boundary rules674.425.04.0 to 5.0
Mutual recognition between schemes across borders664.334.04.0 to 5.0
Public disclosure of the allocation method used674.284.04.0 to 5.0
Physical traceability or chain of custody for the product664.244.04.0 to 5.0
Independent third-party verification674.214.04.0 to 5.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=15Mutual recognition between schemes across borders 4.5
  • Subtask Bn=11Prevention of double counting of emission reductions 4.7
  • Subtask Cn=11Mutual recognition between schemes across borders 4.5
  • Linkage and exchangen=9Transparent eligibility and boundary rules 4.7
  • Industry and policy referencen=15Prevention of double counting of emission reductions 4.6
  • Not yet assignedn=6Public disclosure of the allocation method used 4.7
Q87 Year percentiles n = 62

By what year, if ever, will one steel product carbon-intensity standard, with shared definitions, system boundary and verification, be mutually recognised across the EU, Japan, Korea, the United States, India and China?

Fill the 10% and 90% years first.

10% chance by

2032

IQR 2030–2036 · never: 2

50% chance by

2040

IQR 2035–2048 · never: 4

90% chance by

2045

IQR 2040–2055 · never: 12

0%25%50%75%100%20272031203520392043204720502027: 3% pooled probability2028: 4% pooled probability2029: 7% pooled probability2030: 11% pooled probability2031: 16% pooled probability2032: 22% pooled probability2033: 27% pooled probability2034: 32% pooled probability2035: 37% pooled probability2036: 42% pooled probability2037: 47% pooled probability2038: 51% pooled probability2039: 56% pooled probability2040: 60% pooled probability2041: 63% pooled probability2042: 66% pooled probability2043: 69% pooled probability2044: 71% pooled probability2045: 74% pooled probability2046: 76% pooled probability2047: 78% pooled probability2048: 80% pooled probability2049: 82% pooled probability2050: 84% pooled probability

Pooled cumulative probability: average of each expert's piecewise-linear curve through their 10%, 50% and 90% years.

“After 2050” is counted as 2055 for the statistics
PercentilenMedian yearIQRRange“Never”
10% chance by6020322030 to 20362027 to 20552
50% chance by5820402035 to 20482030 to 20554
90% chance by5020452040 to 20552032 to 205512

By Task 52 affiliation (groups of at least 5)

  • Subtask An=1650% chance by 2040
  • Subtask Bn=1150% chance by 2035
  • Subtask Cn=750% chance by 2035
  • Linkage and exchangen=950% chance by 2040
  • Industry and policy referencen=1650% chance by 2040
Q88 Ranked top 3 n = 52

Rank the three policy instruments most likely to be materially weakened, delayed or reversed by 2030 under political or fiscal pressure.

EU ETS free-allocation phase-out timetable, 2026 to 2034: 60% · 1st 48% EU ETS free-allocation phase-out timetable, 2… 60% · 1st 48% EU CBAM extension to downstream products: 35% · 1st 15% EU CBAM extension to downstream products 35% · 1st 15% US Section 232 steel tariffs at 50%: 35% · 1st 15% US Section 232 steel tariffs at 50% 35% · 1st 15% India's green steel taxonomy procurement mandate: 16% · 1st 6% India's green steel taxonomy procurement mand… 16% · 1st 6% Germany's carbon contracts for difference programme: 14% · 1st 4% Germany's carbon contracts for difference pro… 14% · 1st 4% UK CBAM start in 2027: 11% · 1st 0% UK CBAM start in 2027 11% · 1st 0% Japan's GX-ETS mandatory phase from 2026: 10% · 1st 6% Japan's GX-ETS mandatory phase from 2026 10% · 1st 6% Korea's K-ETS phase 4 steel allocation: 7% · 1st 6% Korea's K-ETS phase 4 steel allocation 7% · 1st 6% China's national ETS steel benchmarks: 6% · 1st 0% China's national ETS steel benchmarks 6% · 1st 0% Australia's Green Iron Investment Fund: 5% · 1st 0% Australia's Green Iron Investment Fund 5% · 1st 0%

Borda score as a share of the maximum (rank 1 = 3 points, rank 3 = 1 point) · share ranking it first.

n = 52 · Borda share is the score as a share of the maximum (rank 1 = 3 points)
OptionBorda shareRanked firstMentioned
EU ETS free-allocation phase-out timetable, 2026 to 203460%48%69%
EU CBAM extension to downstream products35%15%54%
US Section 232 steel tariffs at 50%35%15%52%
India's green steel taxonomy procurement mandate16%6%27%
Germany's carbon contracts for difference programme14%4%25%
UK CBAM start in 202711%0%27%
Japan's GX-ETS mandatory phase from 202610%6%15%
Korea's K-ETS phase 4 steel allocation7%6%10%
China's national ETS steel benchmarks6%0%12%
Australia's Green Iron Investment Fund5%0%10%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=8EU ETS free-allocation phase-out timetable, 2026 to 2034 (Borda 58%)
  • Subtask Bn=11EU CBAM extension to downstream products (Borda 48%)
  • Subtask Cn=9EU ETS free-allocation phase-out timetable, 2026 to 2034 (Borda 70%)
  • Linkage and exchangen=9EU ETS free-allocation phase-out timetable, 2026 to 2034 (Borda 78%)
  • Industry and policy referencen=13EU ETS free-allocation phase-out timetable, 2026 to 2034 (Borda 79%)
Q89 Single choice n = 53

From 2026 CBAM treats imported DRI and HBI as reportable precursors, mostly on verified actual emissions data. How will this affect the competitiveness of imported hydrogen-based iron against EU-produced hydrogen DRI by 2030?

Materially favours EU production; the compliance burden is a real barrier: 19% (10) Materially favours EU production; the complia… 19% (10) Roughly neutral; well-documented low-carbon exporters are not disadvantaged: 47% (25) Roughly neutral; well-documented low-carbon e… 47% (25) Materially favours verified low-carbon imports over EU producers on higher-carbon grids: 15% (8) Materially favours verified low-carbon import… 15% (8) Too uncertain; verification capacity will decide case by case: 19% (10) Too uncertain; verification capacity will dec… 19% (10)
n = 53
OptionExpertsShare
Materially favours EU production; the compliance burden is a real barrier1019%
Roughly neutral; well-documented low-carbon exporters are not disadvantaged2547%
Materially favours verified low-carbon imports over EU producers on higher-carbon grids815%
Too uncertain; verification capacity will decide case by case1019%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=12Materially favours EU production; the compliance burden is a real barrier 33%
  • Subtask Bn=8Roughly neutral; well-documented low-carbon exporters are not disadvantaged 50%
  • Subtask Cn=7Roughly neutral; well-documented low-carbon exporters are not disadvantaged 43%
  • Linkage and exchangen=9Materially favours verified low-carbon imports over EU producers on higher-carbon grids 33%
  • Industry and policy referencen=15Roughly neutral; well-documented low-carbon exporters are not disadvantaged 67%
Q90 Multiple choice, up to 3 n = 54

Under what conditions are carbon contracts for difference most effective in supporting delivery of hydrogen DRI projects? Select up to three.

Coverage of both the carbon-price gap and the hydrogen or electricity cost gap: 76% Coverage of both the carbon-price gap and the… 76% Contract duration of ten years or more: 61% Contract duration of ten years or more 61% Combination with demand-side measures such as lead markets or quotas: 52% Combination with demand-side measures such as… 52% A predictable carbon price and free-allocation phase-out path: 44% A predictable carbon price and free-allocatio… 44% Pairing with grid-connection and hydrogen-supply guarantees: 30% Pairing with grid-connection and hydrogen-sup… 30% Eligibility open to phased, gas-first plants: 20% Eligibility open to phased, gas-first plants 20% Fast award with limited conditionality: 2% Fast award with limited conditionality 2%

Share of respondents selecting each option.

n = 54 · each expert could select up to 3
OptionExpertsShare selecting
Coverage of both the carbon-price gap and the hydrogen or electricity cost gap4176%
Contract duration of ten years or more3361%
Combination with demand-side measures such as lead markets or quotas2852%
A predictable carbon price and free-allocation phase-out path2444%
Pairing with grid-connection and hydrogen-supply guarantees1630%
Eligibility open to phased, gas-first plants1120%
Fast award with limited conditionality12%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=12Coverage of both the carbon-price gap and the hydrogen or electricity cost gap 67%
  • Subtask Bn=11Coverage of both the carbon-price gap and the hydrogen or electricity cost gap 82%
  • Subtask Cn=7Contract duration of ten years or more 71%
  • Linkage and exchangen=8Coverage of both the carbon-price gap and the hydrogen or electricity cost gap 63%
  • Industry and policy referencen=14Coverage of both the carbon-price gap and the hydrogen or electricity cost gap 93%
Q91 Low, best and high estimate n = 21

If the EU's hourly renewable-matching requirement for RFNBO hydrogen takes effect as legislated, by what percentage would it raise the levelised cost of hydrogen delivered to a DRI plant at a typical European site in 2032, compared with annual matching?

Industry estimates range from a few percent to about one third; five member states are seeking a delay to 2035.

Median low

8.0

Median best

20.0

IQR 16.0–35.0

Median high

35.0

020406080100Interquartile range of best estimates: 16.0–35.0Median best estimate 20.0Expert: low 0, best 5, high 15Expert: low 4, best 10, high 21Expert: low 5, best 10, high 20Expert: low 5, best 15, high 40Expert: low 5, best 15, high 35Expert: low 5, best 16, high 26Expert: low 8, best 18, high 33Expert: low 1, best 20, high 20Expert: low 5, best 20, high 35Expert: low 10, best 20, high 30Expert: low 10, best 20, high 30Expert: low 0, best 21, high 35Expert: low 5, best 25, high 35Expert: low 10, best 25, high 40Expert: low 20, best 30, high 100Expert: low 25, best 35, high 50Expert: low 30, best 35, high 40Expert: low 20, best 40, high 60Expert: low 30, best 50, high 60Expert: low 50, best 65, high 100Expert: low 50, best 80, high 100

One line per expert (lowest to highest plausible), dot: best estimate, sorted by best estimate · shaded band: interquartile range of best estimates · dashed: median

Unit: % increase in delivered hydrogen cost

Unit: % increase in delivered hydrogen cost
EstimatenMedianIQRRangeMean
Lowest plausible2185 to 200 to 5014.2
Best estimate212016 to 355 to 8027.4
Highest plausible213530 to 5015 to 10044.0

By Task 52 affiliation (groups of at least 5)

  • Subtask An=8best 25% increase in delivered hydrogen cost
  • Subtask Bn=5best 20% increase in delivered hydrogen cost
Q92 Single choice n = 50

Steel entered China's national emissions trading scheme in 2025 with intensity-based free allocation. How will this affect Chinese steel-sector emissions by 2030?

Material reduction; the scheme meaningfully cuts sector emissions: 8% (4) Material reduction; the scheme meaningfully c… 8% (4) Modest reduction, mainly efficiency gains with little route switching: 52% (26) Modest reduction, mainly efficiency gains wit… 52% (26) Negligible; intensity-based free allocation avoids most real cost: 20% (10) Negligible; intensity-based free allocation a… 20% (10) Too early to judge from the first compliance cycle: 20% (10) Too early to judge from the first compliance … 20% (10)

Mean score 1.9, median 2.0

n = 50 · mean score 1.85, median 2.0
OptionExpertsShare
Material reduction; the scheme meaningfully cuts sector emissions48%
Modest reduction, mainly efficiency gains with little route switching2652%
Negligible; intensity-based free allocation avoids most real cost1020%
Too early to judge from the first compliance cycle1020%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=13Modest reduction, mainly efficiency gains with little route switching 54%
  • Subtask Bn=11Modest reduction, mainly efficiency gains with little route switching 64%
  • Subtask Cn=6Modest reduction, mainly efficiency gains with little route switching 33%
  • Linkage and exchangen=7Modest reduction, mainly efficiency gains with little route switching 86%
  • Industry and policy referencen=10Negligible; intensity-based free allocation avoids most real cost 40%
Q93 Choice grid n = 55

By 2040, what requirements do you expect to apply when an existing blast furnace seeks permission to reline, by jurisdiction?

Choose the most likely requirement in each jurisdiction. Leave a row as not my expertise if you have no view.

European Union European Union · No specific climate requirement: 2% (n=51)European Union · An emissions-reduction plan or abatement-readiness condition: 14% (n=51)European Union · Abatement (CCUS or hydrogen injection) required at reline: 33% (n=51)European Union · Relining of unabated furnaces not permitted: 51% (n=51) Mean score (n=51)2.3 China China · No specific climate requirement: 10% (n=48)China · An emissions-reduction plan or abatement-readiness condition: 46% (n=48)China · Abatement (CCUS or hydrogen injection) required at reline: 33% (n=48)China · Relining of unabated furnaces not permitted: 10% (n=48) Mean score (n=48)1.4 India India · No specific climate requirement: 47% (n=49)India · An emissions-reduction plan or abatement-readiness condition: 47% (n=49)India · Abatement (CCUS or hydrogen injection) required at reline: 4% (n=49)India · Relining of unabated furnaces not permitted: 2% (n=49) Mean score (n=49)0.6 United States United States · No specific climate requirement: 36% (n=47)United States · An emissions-reduction plan or abatement-readiness condition: 40% (n=47)United States · Abatement (CCUS or hydrogen injection) required at reline: 13% (n=47)United States · Relining of unabated furnaces not permitted: 11% (n=47) Mean score (n=47)1.0 Japan Japan · No specific climate requirement: 4% (n=45)Japan · An emissions-reduction plan or abatement-readiness condition: 40% (n=45)Japan · Abatement (CCUS or hydrogen injection) required at reline: 47% (n=45)Japan · Relining of unabated furnaces not permitted: 9% (n=45) Mean score (n=45)1.6 Korea Korea · No specific climate requirement: 4% (n=46)Korea · An emissions-reduction plan or abatement-readiness condition: 39% (n=46)Korea · Abatement (CCUS or hydrogen injection) required at reline: 43% (n=46)Korea · Relining of unabated furnaces not permitted: 13% (n=46) Mean score (n=46)1.7 Other major steel-producing countries Other major steel-producing countries · No specific climate requirement: 29% (n=42)Other major steel-producing countries · An emissions-reduction plan or abatement-readiness condition: 60% (n=42)Other major steel-producing countries · Abatement (CCUS or hydrogen injection) required at reline: 10% (n=42)Other major steel-producing countries · Relining of unabated furnaces not permitted: 2% (n=42) Mean score (n=42)0.9
No specific climate requirementAn emissions-reduction plan or abatement-readiness conditionAbatement (CCUS or hydrogen injection) required at relineRelining of unabated furnaces not permittedNumber at right: mean score
Share of the experts answering each row
RownNo specific climate requirementAn emissions-reduction plan or abatement-readiness conditionAbatement (CCUS or hydrogen injection) required at relineRelining of unabated furnaces not permitted
European Union512%14%33%51%
China4810%46%33%10%
India4947%47%4%2%
United States4736%40%13%11%
Japan454%40%47%9%
Korea464%39%43%13%
Other major steel-producing countries4229%60%10%2%

By Task 52 affiliation (groups of at least 5)

  • Subtask An=15European Union: Relining of unabated furnaces not permitted 40%
  • Subtask Bn=11European Union: Relining of unabated furnaces not permitted 60%
  • Subtask Cn=8European Union: Relining of unabated furnaces not permitted 63%
  • Linkage and exchangen=9European Union: Relining of unabated furnaces not permitted 71%
  • Industry and policy referencen=12European Union: Abatement (CCUS or hydrogen injection) required at reline 55%

IEA Hydrogen TCP Task 52, Hydrogen for Iron and Steelmaking. Expert elicitation 2026, 104 experts, data as of 8 November 2026. Results are pooled expert judgements, not measurements.