Zero Carbon Industrial Systems

Heavy industry emits carbon in two chemically distinct ways, and the distinction decides everything downstream. Energy emissions come from burning fuel to make heat, and can in principle be electrified. Process emissions come from the reaction itself. Conflating them is the source of most confused commentary about this sector, and the confusion is not marginal: it determines whether a technology exists.

For steel the reducing agent is carbon, and carbon can be substituted. Hydrogen strips oxygen from iron ore in place of coke, the resulting sponge iron feeds an electric arc furnace, and the chemistry is not in dispute. There is no analogous substitution in a lime kiln. Calcium carbonate releases CO2 when it decomposes and no heat source changes that. About 60% of cement industry CO2 is process emissions from calcining carbonate and about 40% is fossil energy input; global cement process emissions were 1.58 Gt in 2023, with 46.07 Gt cumulative since 1928. Together steel and cement are about 14% of global energy and process-related CO2 on a direct basis.

The framing under test is that heavy industry can be decarbonised with technology that exists. For steel that is essentially right and misleading in emphasis; for cement it fails, and it fails on chemistry rather than on price. The technology for steel exists, works, and is not being built — because what does not exist is hydrogen at a price the process can bear. For cement, the route that eliminates process emissions requires capturing and burying CO2, the world's one industrial-scale cement capture plant covers half of one plant, and the cheap route that does work is stuck at a standards committee. Both halves of that are content, and neither is a technology gap.

Established The deployment number is the one a page about deployment should carry, and it is devastating. The Green Steel Tracker maintained by the Leadership Group for Industry Transition, with the Stockholm Environment Institute as secretariat, records capacity operational now at 270,000 t/yr of green iron and 60,000 t/yr of green steel. Against world production near 1,800 Mt/yr, the steel figure is about 0.003%. Established Planned capacity for 2050 is 28 Mt/yr of primary green steel plus 18 Mt/yr of green iron — under 2% of today's output as an aspiration for a quarter-century hence. Established And the pipeline has stopped filling: new project announcements peaked at 15 in 2021 and fell to 2 in 2025. Frontier A pipeline that stops being replenished while its front end has not yet delivered is the shape of a sector that has discovered its cost problem — and this tracker is run by an organisation founded to promote industry transition, so the direction of its interest runs against the finding.

Established The flagship is being built and it needed rescuing at 60% complete. Stegra at Boden, Sweden targets 2.5 Mt/yr initial capacity on roughly €7.9 billion raised in total, after a €1.4 billion rescue round led by the Wallenberg family; the plant was about 60% complete in autumn 2025, its original late-2026 target was under review as of April 2026, and the chief executive states it will take 18 to 24 months to produce steel once the facility is finished. Frontier That it survived what trade coverage described as recurrent proximity to collapse is genuine evidence the technology attracts capital; that it needed a rescue at 60% completion is evidence about the cost of first-of-a-kind. Established The other Scandinavian anchor has slipped: SSAB's HYBRIT line ran a 9 kt/yr sponge-iron pilot at Luleå from 2020, two 1.25 Mt/yr electric arc furnaces are now expected in 2029, blast-furnace conversion is a 2030–2040 proposition, mini-mill commissioning moved from 2028 to 2029, and the Oxelösund furnace was separately delayed on permitting in May 2026.

Established The binding number is a hydrogen price, and the gap is a factor of two to three. European hydrogen costs about €5–8 per kilogram; steelmakers need about €2.50–3.00 per kilogram for hydrogen direct reduction to be viable. A single 2 Mt/yr DRI module consumes 140,000–150,000 tonnes of hydrogen a year, European steel hydrogen demand could reach 0.5 Mt by 2030, and only 2.7 Mt/yr of low-emissions hydrogen capacity worldwide has reached final investment decision — for all uses combined. Frontier The industry's response has been to build flexible DRI-EAF plants that run on natural gas and can switch to hydrogen later. That is a rational hedge, a real improvement on a blast furnace, and not the technology the sector is credited with deploying. Whether any plant built that way ever switches is the open question, and no plant has yet.

Established The failure record is the most informative body of evidence in the subject, and the sharpest case is a company that kept its money and changed the project. Cleveland-Cliffs held a $500 million Department of Energy Industrial Demonstrations award against $1.3 billion of its own investment to replace a coal blast furnace at Middletown, Ohio with a hydrogen-capable DRI plant — over 90% lower carbon intensity on renewable hydrogen, about 1 Mt/yr of emissions removed by 2029. The chief executive cancelled it in terms: “It's clear by now that we will not have availability of hydrogen. So, there is no point in pursuing something that we know for sure that's not going to happen.” Established The replacement project relines the 73-year-old No. 3 blast furnace to extend coal use and adds a 70 MW cogeneration plant, with an air permit filed with Ohio EPA in February 2026 and construction starting that September. Frontier A federal decarbonisation award converted into a blast-furnace life extension is the most complete failure case in the sector, and it turns on hydrogen availability rather than on steel technology.

Established Three more losses, and a caution about how to read the tally they sit in. Sublime Systems — whose ambient-temperature electrochemical cement is the most technically radical route in the subject — lost an $87 million DOE award and laid off roughly two-thirds of its workforce. Heidelberg Materials' Mitchell, Indiana capture project, which would have captured about 2 Mt CO2 a year, was cancelled in May 2025. These sit inside a larger withdrawal of 321 awards cancelled in October 2025, 356 across six DOE offices since January 2025, affecting $12.5 billion, with 164 frozen projects unpaid since 1 October 2025. Frontier The source for that tally is an advocacy organisation opposed to the cancellations and should be read as such; the individual cancellations are corroborated by trade and company reporting. Established The European picture is a pause rather than a cancellation wave: Salzgitter delayed its green hydrogen expansion by three years, Shell cancelled the Aukra hydrogen hub in Norway, and Equinor retreated from Norwegian hydrogen projects. Frontier The strongest signal in all of it is not the funding losses but the withdrawals from money already held — a party walking away from a subsidy it won is telling you something a party losing one is not.

Established Cement is the harder half and nothing operating today is at scale. The world's only industrial-scale cement carbon capture plant is Brevik in Norway, inaugurated in mid-2025, capturing 400,000 t CO2 a year — about half the emissions of that one plant — by amine absorption driven by recovered waste heat, at roughly 99% CO2 purity, with the first 1,000 tonnes stored in May 2025. Frontier Set that against the sector: 400,000 tonnes against 1.58 Gt of global cement process emissions is 0.025%. One plant, half-captured, is one four-thousandth of the problem. Established European cement capture installations are costed at €200–500 million per plant, against roughly three thousand cement plants worldwide. Frontier And a capital-allocation signal that is hard to read charitably: Heidelberg Materials' 2024 share buyback of €1.2 billion was roughly three times the cost of its operational capture plant.

Established The clinker-substitution route works now, is cheaper, and is blocked one level below where everyone looks. Limestone calcined clay cement, LC3-50, is about 50% clinker, 30% calcined clay, 15% limestone and 5% gypsum, delivers roughly 40% lower CO2 than ordinary Portland cement at up to 25% lower cost, with better chloride and sulfate resistance, and is in commercial production at 450,000 t/yr in Colombia and 500,000 t/yr in France, plus Ghana, Cuba, Malawi and Denmark. Established EN 197-5 (2021) admits it under CEM II/C-M but permits only ternary blends and caps clinker replacement at 50% even where less would perform; ASTM C1157 is performance-based and does not restrict the number or type of clinker additions. Frontier The decisive gap is one level down: EN 206, the concrete standard, does not yet recognise CEM II/C — and therefore LC3-50 — as a constituent in concrete mixes, and prescriptive minimum cement contents obstruct low-clinker formulations. A cement standard that permits a binder the concrete standard will not accept is not permission to build with it. Established UNIDO's position generalises it: prescriptive standards based on fixed input compositions are the obstacle, and performance-based standards assessing durability, strength and emissions are the remedy.

Frontier One large accounting item cuts the other way and is missing from most sectoral tallies. Concrete reabsorbs CO2 through carbonation over its service life and beyond: uptake reached 0.84 Gt CO2 in 2023, and cumulative uptake of 21.26 Gt offsets roughly 46% of cumulative process emissions since 1928. Established That is a peer-reviewed sink of the same order as the sector's annual emissions. Frontier It makes the cement problem somewhat smaller than the gross figure implies and considerably harder to state honestly, because a net figure invites the inference that the sink can be managed as a lever — and the carbonation happening today is reabsorbing carbon released by concrete poured decades ago.

Established The green premium is measured, small where it matters, and not being paid. The IEA puts near-zero emissions production at 10% to 125% higher cost than conventional, while noting that because these materials are a modest share of end products, the premium on a car or a house is up to 5% and often less. Committed demand against that is thin: publicly disclosed offtakes near 2 Mt/yr by 2030 plus about 3 Mt pledged through buyer coalitions, together worth roughly $3.5 billion; announced 2030 near-zero capacity is about 10% of what a net-zero pathway requires. Established Stated willingness to pay exceeds revealed willingness by a measurable margin. A consultancy survey of steel customers found 42% willing to pay extra if supply were tight by 2030 against 44% who would switch to grey, with under 15% across all materials willing to pay around a 10% premium. Established Market prices are the harder evidence: a price reporting agency assessed the Northern European green flat-steel premium at €100–170/t in January 2026, the Chinese differential at 0–500 yuan/t unchanged since June 2025, and the United States differential at $0 per short ton, flat since the assessment launched in May 2024. Frontier A price reporting agency sells the existence of a market; that its US green steel premium has been zero for over two years is a finding against its own commercial interest. Established A separate trade survey found about 40% of respondents trading or planning to trade green steel, flat products at 71% of green sales against rebar at 7%, construction demand described as “very low”, and one respondent stating that green steel “was a hot topic not long ago, but due to difficult market conditions and high costs, it's been pushed down the priority list”.

Frontier The most uncomfortable open question is whether the premium buys anything at all. Johansson and Kriström argue that within the EU emissions trading system, if the permit supply is exogenous, green steel displacing conventional European production leaves total emissions unchanged — the freed permits are simply used by someone else. Their plant case requires a 45% premium to break even absent cheaper electricity, and they conclude green steel is better sited where electricity is genuinely cheap, outside cap-and-trade coverage. Frontier The argument depends on the permit supply being fixed, and the EU has since introduced a market stability reserve and cancellation mechanisms that weaken it. It is carried here as a live dispute rather than a settled result, because it is the strongest available challenge to the entire voluntary-premium theory of change and it has been contested rather than refuted.

Frontier The steel hypothesis space is unusually well separated by what each route has actually made. Established Hydrogen DRI plus electric arc furnace is established on chemistry — metallisation above 90%, demonstrated — and frontier on commercial viability; what settles it is Stegra producing steel and selling it at a price that covers hydrogen at delivered cost. Established Natural-gas DRI with a hydrogen upgrade path is the dominant actual investment and a genuine improvement on blast furnaces; Speculative that any plant built this way ever switches is an assumption with no instance behind it. Frontier Molten oxide electrolysis made about one tonne of steel from the largest pilot cell run in early 2025 at roughly 1,600 °C, on more than $500 million raised with Tata Steel among the investors — and its developer has since prioritised niobium, tantalum and nickel over steel. A company's own reprioritisation is better evidence than any date it publishes. Speculative Low-temperature electrolytic iron is thinner in public than molten oxide electrolysis and awaits a continuous cell at tonne scale with published energy intensity. Speculative Blast furnace with carbon capture is not effectively demonstrated: the emission points are plural and flue-gas CO2 concentration varies. Established And scrap-based electric arc steel is the largest actually-deployed low-carbon route in the world, at well under a tonne of CO2 per tonne of steel in all-EAF mills against a 1.9 t global average — systematically under-credited because it is not new, and capped below primary demand by scrap availability.

Frontier The cement hypothesis space splits on whether a route touches calcination at all. Frontier Kiln capture has Brevik operating at 400 kt/yr; what would settle it is a second and third plant financed without state capital, and storage capacity to receive them. Established Clinker substitution is established on performance and frontier on adoption, with the barrier named and closeable. Frontier Novel electrochemical binders — ambient-temperature production from non-carbonate feedstocks, eliminating both kiln fuel and calcination CO2 — hold ASTM C1157 General Use designation, the performance-based route that lets a novel chemistry into building codes at all; Speculative their leading developer then lost $87 million and two-thirds of its staff, which is the sector's problem in one company. Speculative Alternative chemistries — belite, calcium sulfoaluminate, magnesium-based and carbonatable binders — are real chemistry, decades old in some cases, blocked by prescriptive standards and by the fact that a building must stand for a century. Frontier Carbon-cured concrete is chemically sound, since carbonation is a measured 0.84 Gt/yr natural sink; Handwave any claim that curing-stage CO2 injection makes concrete carbon-negative is arithmetic nobody has shown, because the quantities injectable are small against the cement's own emissions.

Speculative And a coupling between the two programmes that runs the wrong way. Blast furnace slag is the workhorse supplementary cementitious material, and it becomes scarce as steel decarbonisation removes blast furnaces. Success on the steel side subtracts from the cheapest lever on the cement side. No published pathway this brief located prices that interaction.

Frontier Two hypotheses have almost no constituency and deserve one. Material efficiency — using less steel and cement per unit of service — rests on a substantial engineering literature showing over-specification in construction is large and measurable, and has no vendor, which is why it is the most under-resourced live option in the subject. Frontier And industrial heat storage, which is the strongest single support for the slot's framing anywhere in the evidence: a brick-and-wire heat battery reached 100 MWh at a Californian fuel-production site in October 2025, delivering steam at over 1,000 °C with round-trip efficiency above 97%, on off-grid solar, requiring no air permits, after ten weeks of automatic operation. Frontier It stores heat to deliver heat, avoids two conversions, uses materials with no supply constraint and displaces gas boilers directly. It is also a company release with no cost figure, and 100 MWh is one installation. Energy Storage Revolutions reaches the same conclusion from the storage side, and this brief defers to it on duration, dispatch and market design.

Established First and largest: the cost differential under global competition. A producer using clean processes competes against one that is not, in a commodity market with global overcapacity, and no technical advance closes that by itself. The evidence that it binds is direct: the US green steel premium at zero for over two years, the European premium at €100–170/t against a cement cost premium above 100%, and buyers' stated ceiling at about 10%.

Established Second: input price, specifically hydrogen and electricity. The €5–8/kg against €2.50–3.00/kg gap is the proximate cause of every steel withdrawal in the record, including the one where the chief executive said so in terms. Frontier This is not a technology gap and it is important not to file it as one. What steel needs is electrolyser capacity and a power price — an industrial fact about how much equipment exists and what electricity costs, not a claim that hydrogen is the right energy carrier. Hydrogen Economies adjudicated the carrier thesis as failing on storage, heating and transport and holding on industrial feedstock substitution, which is the leg this brief sits on; the evidence assembled here sharpens rather than contradicts that finding, because even the surviving leg is not converting at present prices.

Established Third, and for cement only: standards. EN 206 not recognising CEM II/C is a specific, nameable, closeable obstruction that no amount of cement-plant investment routes around. Frontier It is also a fact about a standards body rather than about a technology, which is why it is recorded below as an institutional requirement. A committee could close it; a laboratory cannot.

Frontier Fourth: capital stock turnover. Blast furnaces last decades and the replacement decision is rare. Cleveland-Cliffs relining a 73-year-old furnace locks in coal for another campaign, and that decision was made in 2026 — which means the next opportunity at that site is a generation away. Every year in which a reline happens instead of a rebuild removes a decision point rather than delaying one.

Speculative Fifth, and least discussed: CO2 transport and storage. Every cement pathway that survives contact with process chemistry ends in geological storage, and cement capture at scale is a pipeline-and-reservoir problem rather than a cement problem. Established This brief covers capture on a kiln — capture rate, energy penalty, cost per plant, count of operating installations — and hands transport, reservoir capacity and permanence to the carbon programme's capture treatment rather than re-arguing them.

Established Nothing on this map produces a result this brief waits on, and that is a stronger statement here than in most subjects. Nothing in the evidence base waits on a scientific discovery. Steel waits on an input price; cement waits on chemistry that has no substitute plus a standards code; both wait on a buyer. No typed depends-on edge is claimed.

Established What the brief waits on from research is narrow and mostly measurement rather than discovery. A published delivered cost per thermal MWh for industrial heat batteries against gas, which no vendor release gives. A continuously operating molten oxide electrolysis cell with published energy intensity. A durability dataset for LC3-50 concrete long enough to satisfy a code committee, which is the actual content of the EN 206 obstruction. And a lifecycle accounting convention for carbonation uptake that neither ignores 0.84 Gt/yr nor lets it be claimed as an offset. Frontier The most consequential missing number is not technical at all: nobody has published what a second and third cement capture plant would cost without state capital, and the first one was built with it.

Established The cheapest high-value experiment is a standards trial, not a laboratory one. EN 206 admitting CEM II/C requires durability and strength evidence on LC3-50 concrete over the periods a code committee accepts. That evidence is generatable now, on a binder already in commercial production at 450–500 kt/yr in two countries, with measured chloride and sulfate advantages already in hand. Frontier It is the most actionable finding in the subject and it is bounded by committee calendars rather than by science.

Frontier Second: run the switch. The flexible DRI-EAF hedge rests on an untested assumption that a gas-fired plant converts to hydrogen when the price arrives. No plant has done it. A single documented conversion — with the capital cost, the downtime, the metallurgical differences and the resulting delivered cost published — would convert the sector's dominant investment thesis from an assertion into a datum, in either direction.

Established Third: publish the delivered cost of industrial heat. A heat battery at 100 MWh, over 1,000 °C and above 97% round-trip is the strongest support the framing has, and its cost per thermal megawatt-hour against gas has not been published. Frontier Until it is, the strongest case for the slot's own framing rests on a vendor release.

Frontier Fourth: settle the permit-displacement argument empirically rather than theoretically. The claim that a green premium inside a fixed emissions cap changes nothing is testable against the EU's post-reform permit supply, where the market stability reserve and cancellation mechanisms are exactly the parameters that decide it. Speculative Nobody has published that test, and the theoretical dispute has been running since 2022 while billions are allocated on the assumption that it is wrong.

Established Where “technology that exists” is most true and least true is a spectrum, not a binary, and the chemicals sector shows both ends. Electrified steam cracking is demonstrated: BASF, SABIC and Linde started the world's first large-scale electrically heated cracking furnace in April 2024 at 6 MW, processing about 4 tonnes an hour of hydrocarbon feedstock, claiming at least 90% lower emissions, on €14.8 million of German federal funding. Frontier It is a demonstration still comparing two heating approaches to determine which scales, and 4 t/h is roughly an order of magnitude below a world-scale cracker. Calling that “technology that exists” is defensible for the chemistry and premature for the plant.

Established Ammonia is the clearest case where the technology exists and the economics do not. Conventional ammonia consumes about 5% of global natural gas. A 2026 optimisation of decentralised electric Haber–Bosch finds 2025 costs exceeding market prices by more than $500 per tonne in regions with limited renewables, improving by 2045 to viability in the United States and Ethiopia with premiums of $175–435 per tonne remaining in China, Brazil and India. Frontier A twenty-year path to parity in two countries is a real modelling result and it is not a deployable technology today.

Established The engineering constraint on the steel route is not the furnace but the electrolyser fleet upstream of it. One 2 Mt/yr DRI module wants 140,000–150,000 tonnes of hydrogen a year. Against that, 2.7 Mt/yr of low-emissions hydrogen capacity worldwide has reached final investment decision for all uses. Frontier Roughly eighteen modules would consume the entire committed global low-carbon hydrogen supply, and Europe alone plans more steel capacity than that. This is a manufacturing-capacity number, and it is why the requirement recorded below is industrial rather than scientific.

Established On the cement side the engineering question is an energy penalty and a purity spec. Brevik's amine capture is driven by recovered waste heat rather than by additional fuel, which is what makes 400,000 t/yr at roughly 99% purity achievable on a kiln at all, and it captures about half the plant. Frontier Whether the other half is capturable at acceptable energy cost, at three thousand plants with different heat balances, is not established by one installation — and the €200–500 million per-plant cost is a European figure for European plants.

Within this map: Hydrogen Economies, which owns hydrogen production economics, electrolyser costs, the auction and withdrawal record, leakage and the carrier claim — this brief reaches hydrogen only as a delivered input price and does not restate its numbers as findings of its own; Energy Storage Revolutions, which names thermal storage for industrial heat as the most promising non-electric route and owns duration, dispatch and market design; Advanced Fission, where high-temperature process heat is a stated prize for gas-cooled reactors and the one operating high-temperature reactor retreated to a steam cycle; Advanced Battery Technologies and Geothermal Megaprojects, as the electricity-side and heat-side supply options a green industrial site would draw on.

The boundary with Hydrogen Economies is the one that matters and it should be stated precisely. That brief concluded the hydrogen-as-carrier thesis fails on three legs and holds on one — industrial feedstock substitution, where about 60% of all committed low-emissions hydrogen volume is going — and adjudicated itself as depending on nothing on this map. Nothing here contradicts it. The honest framing of the steel route is that what steel needs is electrolyser capacity and a power price, which is an industrial fact rather than the carrier thesis, and that the demand-side ceiling that brief measured is the mechanism behind every withdrawal recorded in section 2.

This brief supersedes, and does not replace, the carbon programme's treatment, which reaches the same subject from the emissions side and remains the shorter route in. The operational tonnage, the failure record, the quantified hydrogen gap, Brevik, the EN 206 obstruction, carbonation uptake, CBAM's cost, ammonia, electric cracking and industrial heat storage are new to this brief.

Outside the map: process metallurgy and cement chemistry; construction codes and standards bodies, which supply the binding institutional constraint; trade policy, which supplies the border adjustment; and materials-flow analysis, which owns scrap availability and the material-efficiency question this brief flags as under-resourced.

Established The EU Carbon Border Adjustment Mechanism entered its definitive period on 1 January 2026, and it is measurable in cost. Importers of covered goods including steel, cement and fertilisers must now purchase certificates rather than merely report, with a 50-tonne annual de minimis per importer excluding hydrogen and electricity, phasing to full implementation by 2034. The Commission set China's default hot-rolled coil emissions value at 3.187 t CO2 per tonne of steel, which a price reporting agency calculates as roughly €145 per tonne of CBAM cost — larger than the entire European green steel premium of €100–170/t. Frontier Whether it has changed anything measurable is genuinely open and this brief declines to claim it has. The definitive regime is months old; the observable effects so far are a cost on importers and an exporter complaint; the effects that would matter — a rise in the premium actually paid, green capacity reaching final investment decision, a measurable shift in import emissions intensity — are not in evidence, and project announcements fell to two in 2025 while the transitional phase ran. Handwave Attributing any 2026 investment decision to CBAM at this date is assertion.

Established The standards system is the sector's most underrated institution and its most specific lever. Prescriptive standards specify fixed input compositions; performance standards specify what the material must do. The difference decides whether a novel binder can be used at all: ASTM C1157 is why an electrochemical cement can enter US building codes, and EN 206 is why a cement the European cement standard already permits cannot yet go into European concrete. Frontier A standards committee is a slow institution with a strong and defensible reason for being slow — a building must stand for a century, and the failure mode of a bad concrete spec is a collapse decades later. That is a legitimate reason for caution and it is still the binding constraint on the cheapest cement lever available.

Established And most of the good evidence here comes from interested parties, in both directions. The green steel tracker is run by an industry-transition advocacy group and reports figures that undercut its own mission. A price reporting agency sells market assessments and reports a US premium of zero for over two years. A consultancy with clients in the sector reports that 44% of steel customers would switch to grey. Frontier The one tally that runs with its author's interest — the DOE cancellation total from an organisation opposed to the cancellations — is flagged as such and its individual entries corroborated elsewhere. Where a finding runs against the publisher's interest, this brief weights it more heavily, and says so.

Established The distributional question in this sector is about who pays and it has a measured answer that is easier than it sounds. The production cost premium is 10–125%; the premium on a finished car or house is up to 5% and often less. Frontier The affordability objection to green materials is therefore much weaker than the commodity-price numbers suggest, and the willingness-to-pay evidence still says buyers do not pay it — under 15% of customers across all materials would accept around 10%. That gap between a small real cost and an absent willingness is the sector's actual ethical problem, and it is a coordination failure rather than a hardship.

Frontier The carbonation finding creates a genuine accounting hazard. Concrete reabsorbs 0.84 Gt CO2 a year, offsetting roughly 46% of cumulative process emissions. Reporting the gross figure overstates the sector's net contribution; reporting the net figure invites a producer to claim a sink created by its predecessors' emissions decades earlier. Speculative Neither convention is honest without the other, and most sectoral accounting simply omits the term. Stating it and refusing to net it is the position this brief takes.

Established Border adjustment is a distributional instrument wearing an environmental one's clothes. A €145 per tonne cost on Chinese hot-rolled coil is larger than the entire European green premium, is paid by importers, and lands hardest on exporters with the least capacity to decarbonise. Frontier The environmental case for it is that it prevents leakage; the measurable effect so far is a cost, and the exporter complaint on the record is that it undermines the competitiveness of the exporter's own green steel. Handwave A mechanism whose fairness case rests on effects that are not yet measurable is being defended on a forecast.

Established The terminal position is a split verdict, and both halves are load-bearing. For steel the framing is essentially right and misleading in emphasis. Hydrogen DRI works, is being built, and produces steel. What does not exist is hydrogen at €2.50–3.00 a kilogram, and in this context hydrogen is not a technology — it is a delivered commodity price. Saying heavy industry can be decarbonised with technology that exists, while the enabling input costs two to three times what the process can bear, is true in the same way that air travel can be decarbonised with existing technology given unlimited synthetic kerosene.

Established For cement the framing fails, and it fails on chemistry rather than price. Sixty per cent of cement emissions are process emissions and no existing technology removes them without capturing and burying the CO2. The world's one industrial-scale cement capture plant covers half of one plant and 0.025% of global cement process emissions, at €200–500 million a plant against roughly three thousand plants. Clinker substitution — the route that works, is cheaper and already produces hundreds of kilotonnes a year — tops out at about 50% clinker replacement under the standard that permits it, and the concrete standard below does not yet recognise the resulting cement at all. Novel binders that would eliminate calcination exist at pilot scale and their leading developer just lost two-thirds of its staff.

Frontier The sharpest single fact in the subject is Cleveland-Cliffs, and it is sharp because nothing about it is a technology failure. A steelmaker holding a $500 million federal decarbonisation award, in a jurisdiction with a hydrogen hub programme, examined the technology that exists and chose to reline a 73-year-old coal blast furnace and add 70 megawatts of cogeneration instead — saying publicly that hydrogen availability was why. That is not a technology gap. It is also not a decarbonisation outcome, and no amount of describing the technology as existing changes what got built.

Frontier What follows is a change of question rather than a verdict on effort. The useful question is not which process to invent but what would have to be true for the existing process to be chosen: an electrolyser fleet an order of magnitude larger than the one that has reached investment decision, a power price that makes it run, a concrete code that admits a binder already in commercial production, and a buyer — public or private — who pays a premium that is under 5% of a finished product and is currently zero in the largest steel market on the record. Speculative None of those is discovered in a laboratory, and treating this as a research subject is how a solved chemistry problem stays an unsolved emissions problem for another decade.

These horizons track plant commissioning dates, standards committee cycles and regulatory phase-ins rather than research results:

• 10 yr: Frontier The dated items resolve in this window: Stegra producing commercial steel on a target already under review, SSAB's two 1.25 Mt/yr electric arc furnaces in 2029, and CBAM phasing to full implementation by 2034. Established On the tracker's own numbers the 2050 aspiration is under 2% of today's output, so nothing in this window changes the sector's emissions materially. Speculative The most consequential possible event is unglamorous and cheap: EN 206 admitting CEM II/C, which would unlock a binder that is already 40% lower in CO2 and up to 25% cheaper, at hundreds of kilotonnes a year of existing production. Frontier Expect industrial heat storage to be the fastest-moving line, because it displaces gas boilers directly, needs no air permit and competes on a fuel price rather than a carbon price.
• 25 yr: Frontier This is the window in which the hydrogen price either falls to €2.50–3.00 a kilogram or the steel route does not happen at scale, and the determinant is electrolyser manufacturing capacity and power prices rather than steelmaking. Speculative Electric Haber–Bosch reaches modelled viability in a small number of favourable regions around 2045, with $175–435 per tonne premiums persisting in the largest consuming countries. Speculative For cement, the plausible split is that clinker substitution carries most of the achievable reduction and kiln capture remains a small number of state-supported plants, because the capture route needs transport and storage that is a different industry's problem.
• 50 yr: Speculative If cement process emissions are eliminated rather than reduced, it happens at this horizon and through a binder chemistry that does not calcine carbonate — the electrochemical and alternative-chemistry routes now at pilot scale. Speculative The blocker at that horizon is a century-old code culture rather than a reaction, and code cultures change on generational timescales for defensible reasons. Handwave Any projection of net-zero heavy industry at this horizon is assuming a standards outcome and a buyer, neither of which is forecastable from current data.
• 100 / 250+ yr: Handwave Beyond useful forecasting. The one structural observation is that the concrete already poured keeps reabsorbing CO2 — 0.84 Gt a year, 21.26 Gt cumulative, about 46% of cumulative process emissions — so the sector's very long-run net footprint is smaller than its gross record by a term nobody manages and few count. Handwave Whether that is a comfort or an accounting artefact depends on a convention that does not yet exist.

• Depends on Nothing on this map. No brief in this corpus produces a result this subject waits on, and the reason is unusually clean: nothing here waits on a scientific discovery at all. Steel waits on a delivered hydrogen price, cement on chemistry that has no substitute plus a concrete code, and both on a buyer. No typed depends-on edge is claimed.
• Requires (not on this map) The first is a fact about a standards body rather than about a technology. EN 197-5 has admitted limestone calcined clay cement under CEM II/C-M since 2021, but EN 206 — the concrete standard one level below — does not yet recognise CEM II/C as a constituent in concrete mixes, and prescriptive minimum cement contents obstruct low-clinker formulations. A cement standard that permits a binder the concrete standard will not accept is not permission to build with it, and LC3-50 is already produced at 450,000 t/yr in Colombia and 500,000 t/yr in France at roughly 40% lower CO2 and up to 25% lower cost. A committee could close this; a laboratory cannot. The second is a price nobody is paying: near-zero production costs 10–125% more, buyers state a ceiling near 10%, under 15% of customers across all materials would accept even that, the Northern European green flat-steel premium is €100–170/t, and the United States differential has been $0 per short ton since the assessment launched in May 2024. The third is manufacturing capacity rather than the hydrogen carrier thesis: a single 2 Mt/yr direct-reduction module consumes 140,000–150,000 tonnes of hydrogen a year, against 2.7 Mt/yr of low-emissions capacity worldwide that has reached final investment decision for all uses combined, and European hydrogen at €5–8/kg against the €2.50–3.00/kg the process can bear. The fourth is the pipeline and reservoir side of cement capture, which is a different industry's asset base: one operating industrial-scale cement capture plant, at 400,000 t/yr covering half of one plant and 0.025% of global cement process emissions, against roughly three thousand plants at €200–500 million each.
• Enables In principle, decarbonised steel and cement are inputs to almost every physical system on this map. No typed enabling edge is claimed, because at 60,000 t/yr of operational green steel — 0.003% of world production — the enabling relation is a plan rather than a supply, and an edge asserted on 0.003% would misrepresent a pipeline whose new announcements fell from 15 in 2021 to 2 in 2025.
• Adjacent Process metallurgy and cement chemistry; construction codes and standards bodies; trade policy, which supplies the border adjustment; materials-flow analysis, which owns scrap availability and material efficiency; and within this map Hydrogen Economies, Energy Storage Revolutions and Advanced Fission.

Handwave “Process emissions can be electrified away.” They cannot, and this is the single most-confused point in public commentary about the sector. Calcining calcium carbonate releases CO2 from the rock regardless of how the kiln is heated. About 60% of cement's emissions are in that category, and the only routes that touch them are capturing and burying the CO2, using less clinker, or using a binder that is not made from carbonate.

Frontier “Green steel is scaling.” Operational capacity is 60,000 t/yr — 0.003% of world production. New project announcements fell from 15 in 2021 to 2 in 2025. The 2050 aspiration is under 2% of today's output. Established A growth rate computed from a base of essentially zero, on a pipeline that has stopped being replenished, is not a scaling story, and the organisation publishing these figures was founded to promote the transition they undercut.

Frontier “The failure record is a policy-reversal story.” Partly, and the more informative cases run the other way. Thyssenkrupp suspended a hydrogen tender on price; ArcelorMittal declined €1.3 billion of German subsidy it had already won; Cleveland-Cliffs kept its $500 million award and changed the project to a blast-furnace reline. Established Parties withdrawing from money they already held is a stronger signal than parties losing funding, and treating the DOE cancellation wave as the explanation obscures it.

Frontier “Carbon capture is a demonstrated route for cement.” One plant, half-captured, built with state support, capturing 0.025% of global cement process emissions. Established Brevik is a real and well-executed installation and the inference from one installation to a sector of three thousand plants at €200–500 million each — each needing somewhere to put the CO2 — is not supported by it.

Speculative “Molten oxide electrolysis is near commercial.” The largest pilot cell run in early 2025 produced about one tonne of steel, and the developer has since prioritised niobium, tantalum and nickel. Frontier A company's own reprioritisation away from steel is better evidence than any date it publishes, and the same discipline applies to electric steam cracking: a 6 MW, 4 t/h demonstration still comparing two heating architectures is not a plant.

Frontier “A voluntary green premium can carry the transition.” About 2 Mt/yr of disclosed offtakes plus 3 Mt pledged, against 1,800 Mt/yr of steel; construction demand described as “very low”; the US premium at zero for over two years. Frontier And a peer-reviewed argument — contested but not refuted — that inside a fixed emissions cap the premium changes nothing, because green steel displacing conventional European production frees permits that someone else uses. Frontier That argument depends on an exogenous permit supply the EU has since modified, which is why it is flagged frontier rather than established. It has not been tested empirically and billions are being allocated on the assumption that it is wrong.

Handwave “CBAM is driving the transition.” The definitive regime began on 1 January 2026 and has produced a measurable cost — roughly €145 per tonne on Chinese hot-rolled coil, larger than the entire European green premium. It has not yet produced a measurable emissions or investment effect, and project announcements fell to two in 2025 while its transitional phase ran. Attributing any 2026 investment decision to it at this date is assertion.

Frontier “Concrete carbonation means cement is nearly carbon-neutral.” Carbonation is real, peer-reviewed and large — 0.84 Gt in 2023, 21.26 Gt cumulative, about 46% of cumulative process emissions — and it is absent from most sectoral accounting. Handwave It is also not a lever: today's uptake is reabsorbing carbon released by concrete poured decades ago, the timescale is a service life and beyond, and any claim that curing-stage CO2 injection makes concrete carbon-negative is arithmetic nobody has published, because the injectable quantities are small against the cement's own emissions.

Established “Scrap-based electric arc steel is not really low-carbon.” It is the largest actually-deployed low-carbon steel route in the world, at well under a tonne of CO2 per tonne of steel in all-electric mills against a 1.9 t global average. Frontier It is under-credited because it is not new, and its real limit is scrap availability, which caps it below primary demand — a materials-flow constraint rather than a technology one.

Speculative And the framing itself deserves one correction that runs against the sector's optimists and its critics at once. Nothing in this evidence base waits on an undiscovered process. Established The chemistry problem for steel is solved, the chemistry problem for cement is not solvable by substitution, and neither of those facts is what determines whether emissions fall. What determines it is a hydrogen price, an electrolyser fleet, a concrete code and a buyer — and describing any of those as a research frontier is a category error that has cost the sector a decade.