The Future of European Steel Under CBAM

Europe's steel future will be decided where five clocks meet: ageing blast furnaces, declining operating output, the EU ETS and CBAM transition, the delivery of electricity and low-carbon hydrogen, and a global market carrying large excess capacity. CBAM can change the relative carbon treatment of covered imports. It cannot by itself build grids, produce suitable scrap, supply direct-reduction pellets, finance a mill conversion or create buyers willing to sign long-term offtake.

The strongest conclusion is therefore conditional: European steel can decarbonise through more than one route, but each route has a different binding input and delivery test. The useful question is not “Will CBAM save steel?” It is “Which evidence would prove that one asset, product and supply chain has a viable transition case?”

This report owns the steel-sector transition map: production structure, routes, asset timing, country differences, input constraints, trade pressure, project delivery states and the monitoring decisions those facts create. It does not duplicate the site's existing canonical answers.

For a product decision, use the CBAM goods checker. This article is a sector map, not a customs-classification ruling.C-001C-002

EUROFER's operating-capacity series falls from 236 Mt in 2008 to 194 Mt in 2025. Its utilisation series moves from 78.4% to 64.8% across the same endpoints, with large cyclical breaks in between. These are industry-reported operating measures, not a universal test of mill viability. But together they show why a capital-heavy route conversion cannot be analysed as a technology exercise alone: the investment case starts inside an already under-used production system.C-014C-015

The decline is not one smooth structural line. Utilisation fell to 55.5% in 2009, recovered above 70% in several years, fell to 62.1% in 2020 and reached 60.1% in 2023. A single endpoint cannot tell whether demand, outages, imports, energy prices or plant decisions caused the movement. It can tell a strategy team which evidence must be added before approving a route narrative.

Germany's 34.090 Mt represented 27.1% of the 2025 EU country-table total. Italy contributed 20.733 Mt, Spain 11.778 Mt and France 9.916 Mt. The table then extends through every printed country row, not a hand-picked top five. That matters because the asset, electricity, scrap and product mix behind one tonne in Germany is not the same as one tonne in Italy, Sweden or Luxembourg.C-012

Country coverage also needs an honest missing-data rule. EUROFER prints 19 named rows and an Others aggregate. Croatia and Hungary display dashes in 2025. This report therefore covers the full published population while refusing to invent seven individual country figures. The atlas is a reference table, not a transition ranking.C-013

The global comparison is deliberately separate. Worldsteel's table contains 50 countries plus Others, led by China at 960.8 Mt and India at 164.9 Mt. Those values describe national crude-steel production—not capacity, exports to Europe, covered CBAM goods or verified embedded emissions. They show the asymmetry of the production field; they do not assign a CBAM risk score.C-020C-021

The route question is often reduced to coal versus hydrogen. That is too crude for an investment, sourcing or evidence decision. Europe's portfolio includes BF-BOF assets, scrap-EAF production, natural-gas DRI-EAF bridges, hydrogen DRI-EAF, capture configurations and early electrolysis pathways. Their inputs, maturity, residual emissions and infrastructure dependencies differ.

JRC reports H₂-DRI-EAF at technology-readiness level 6–8 and cites an average abatement-cost range of EUR 122–171 per tonne CO₂. It also cites scrap-EAF production costs 35–68% above BF-BOF in the underlying comparison. Both ranges are sensitive to electricity, hydrogen, scrap quality and other assumptions. They are not current quotes for a mill or a forecast of a steel premium.C-032C-033

Natural-gas DRI can be a bridge toward hydrogen, and modern DRI plants can be designed as hydrogen-ready. Neither phrase proves that low-carbon hydrogen is being used today. EAF is likewise a furnace route, not a guarantee of 100% scrap, renewable electricity or near-zero product emissions.C-034C-035

JRC describes an EU blast-furnace fleet averaging about 50 years, with 25% operating for more than 60 years. Almost half of EU blast furnaces are set to retire before 2035 in the cited synthesis. That creates a decision window: reline, convert, replace, pause or close. It does not prove which choice any installation will make.C-018C-019

A strategy passes this gate only if the three clocks align. A technically mature process without grid access is not an operating pathway. A subsidised project without financing, construction and offtake is not delivered capacity. A low-emission product without a compatible emissions boundary and buyer acceptance is not automatically a premium market.

JRC reports current EU steel electricity use of 75 TWh per year, including 55 TWh purchased from the grid; the remainder is generated mainly from primary-steelmaking residual gases. Its announced-project pathway then requires 75 TWh per year of direct electricity plus about 2.12 Mt of hydrogen. Producing that hydrogen by water electrolysis corresponds to another 90 TWh, for about 165 TWh per year of fossil-free electricity.C-036C-037C-038

The correct comparison is not 75 + 165. The 165 TWh pathway is the stated 75 TWh direct requirement plus 90 TWh for electrolysis. It is conditional on all announced plants running and on the specified hydrogen-production route. It does not prove that generation, network capacity, electrolysers, storage or contracts exist.

For a project file, replace “hydrogen available” with four fields: delivery point, annual quantity, carbon-accounting method and delivered price period. Replace “renewable electricity secured” with generation profile, grid connection, contractual instrument and treatment under the emissions methodology. Those fields turn a slogan into evidence.

EUROFER reports EU scrap consumption of 74.424 Mt in 2025, imports of 4.797 Mt and exports of 16.309 Mt. Its published net-export value is 11.511 Mt; subtracting the rounded gross values produces a one-thousand-tonne difference. That rounding gap is preserved rather than “corrected”.C-025C-026C-027

Net exports are not an immediately recoverable domestic feedstock pool. Scrap differs by grade, copper and other residuals, collection and sorting cost, recovery yield and product suitability. A flat or automotive grade can impose different input constraints from construction steel. JRC's pathway estimates EU scrap use at about 77 Mt in 2024 and 92 Mt in 2035 under its route-mix and stable-production assumptions; that is a conditional pathway, not proof of a quantified shortage.C-031

Primary routes face their own dependency. JRC estimates 77% EU import reliance for iron ore and reports around 75% of iron-ore and coking-coal supply coming from extra-EU sources. Those values describe supply structure, not interruption probability. A DRI pathway adds a further question: whether the ore or pellet supply has the quality needed by the selected process.C-028C-029

OECD estimates global excess steel capacity at 640 Mt in 2025 and projects up to 745 Mt in 2028. The projected number depends on announced projects underway and planned plus the OECD demand path. It is not an observed 2028 fact. OECD separately reports global steelmaking capacity at a record 2,445 Mt in 2025 while demand contracted for a fourth year.C-040C-041C-042

EUROFER's all-qualities import-origin table totals 29.727 Mt in 2025. Türkiye is the largest named row at 4.815 Mt. Indonesia rises from 0.088 Mt in 2016 to 2.636 Mt in 2025. These endpoints do not prove CBAM causation, trade diversion or supplier emissions; product mix and other trade measures also changed.C-022C-023C-024

Trade protection and carbon adjustment are separate layers. The pre-existing steel safeguard expired after 30 June 2026. Regulation (EU) 2026/1384 opened the replacement tariff quotas; Implementing Regulation (EU) 2026/1457 distributed those quotas and applies from 1 July through 31 December 2026. CBAM uses a different legal mechanism. A commercially useful steel view keeps quota access, duty exposure, CBAM evidence and product demand in separate scenario rows rather than collapsing them into one “border cost”.C-008C-009C-056

Current CBAM scope is narrower than “all steel-containing goods.” Annex I lists covered iron and steel codes and exclusions, including ferrous waste and scrap under CN 7204. The downstream proposal describes around 180 additional CN codes and an estimated 7,500 additional importers, with proposed application from 1 January 2028. Those remain proposal parameters—not current obligations.C-001C-002C-006C-007

CBAM certificate surrender is adjusted to reflect EU ETS free allocation for equivalent EU production, while the ETS Directive applies a declining CBAM factor to free allocation for covered goods. This article does not reproduce the schedule or calculate a company result. Use the dedicated phase-in guide and the CBAM cost calculator with current official assumptions.C-003C-004

The JRC synthesis says most announced EU near-zero projects combine hydrogen-compatible DRI-EAF and scrap-EAF routes. It also states the prerequisite directly: project realisation requires a business case. Therefore announced projects are not operating capacity. Neither aid approval, a company target nor a technology-readiness claim closes the delivery chain.

A May 2026 JRC INCITE report record describes a 2026–2030 pipeline of approximately 17 Mt per year of direct-reduced-iron capacity across eight plants and about 35 Mt per year of new electric-arc-furnace capacity across sixteen plants. The record's use of “confirmed” does not make those tonnes commissioned capacity, annual production or proof that all plants will operate.C-058

JRC cites EUR 31 billion of CAPEX and EUR 54 billion of OPEX—EUR 85 billion combined—for the announced near-zero project pathway to 2030. It separately reports a EUROFER projection of 81.5 MtCO₂ in annual emissions reductions by 2030 from a 145 MtCO₂ 2023 baseline if the projects are completed. These are conditional scenario figures, not audited committed budgets, achieved abatement or a project-by-project forecast.C-039C-057

A project should enter a capacity figure only when the dataset's state and measurement allow it. Announcement counts can support a pipeline map. Commissioned nameplate capacity can support a commissioning table. Observed annual production can support an output table. Those are three different artifacts.

Supply-side technology is only half the market. A route conversion requires products buyers can specify, emissions boundaries they can compare and offtake terms that help support financing. JRC's comparison of international initiatives finds multiple low-carbon-steel definitions rather than one universal threshold. Accounting boundaries, scrap treatment and performance classes differ.C-044

Global industry data allocate 52% of 2025 steel use to building and infrastructure, 16% to automotive, 12% to mechanical equipment, 10% to metal products and smaller shares to other uses. That is a global demand structure—not an EU CBAM allocation. It explains why public procurement, construction standards, vehicle policy and industrial buyers can matter to route economics without every downstream product being currently covered by CBAM.C-046

The fastener case has its own canonical workflow. Buyers of CN 7318 goods should use the steel-fastener buyer guide. The proposed broader downstream extension is covered by the current-versus-proposed downstream guide.

The research cut-off is 12 August 2026. The source search covered current EUR-Lex law and legislative procedure, Commission steel and CBAM documents, JRC route, supply-chain, intensity and definition studies, OECD capacity analysis, worldsteel and EUROFER statistical publications, IEA technology analysis and official operational pages. Primary law owns legal claims. Institutional studies own pathway claims. Industry publications are labelled rather than presented as neutral official statistics.

1. Freeze: material PDFs and page text were retained locally with checksums where retrieval allowed.
2. Classify: adopted law, official operations, proposal, institutional pathway, industry data and held conclusion remain separate.
3. Atomise: every material claim records a source, locator, limitation, prohibited inference and canonical owner.
4. Reconcile: complete printed populations are reproduced; dashes, aggregates and rounding differences remain visible.
5. Fail closed: unsupported shortages, project delivery, company outcomes and universal technology conclusions stay in the held ledger.

No chart mixes country output, capacity, trade or embedded emissions. The EU table covers 19 named rows plus Others. The world table covers 50 countries plus Others. OECD's 2025 and 2028 excess-capacity values retain estimate and projection labels. JRC's project energy stack retains its “all announced plants” and electrolysis conditions.