AISI Backs EPA's Power-Plant Carbon Rollback — But the Relief Is Narrower Than the Headline Suggests

AISI's endorsement of EPA's power-plant carbon rollback is a straightforward trade-association intervention, but it repays reading closely for what it does and does not say about the US steel industry's electricity problem.

On 14 September 2026, Kevin Dempsey, AISI president and CEO, commended EPA for repealing the 2024 greenhouse gas standards for power plants, arguing that the prior limits rested on carbon capture and sequestration (CCS) delivering at least 90% emissions reductions — a system "not adequately demonstrated to be commercially available for widespread use in the power sector." The relief, in AISI's framing, is reliable and affordable electricity for steel producers, and the ability to "continue to lead the world in low emissions and energy efficient steel production."

The substance of AISI's argument tracks the regulator's own reasoning almost word for word. The economically significant part of the story for metals readers, though, is not the repeal itself but the electricity demand curve that sits behind it, and the regulatory durability that AISI's statement glosses over.

What EPA actually did — and did not do

AISI's statement describes a repeal of "the 2024 greenhouse gas emissions standards for power plants." The final rule is narrower. The Federal Register text is titled Partial Repeal of the Carbon Pollution Standards for Fossil Fuel-Fired Electric Generating Units (91 FR 58954, published 17 September 2026, effective 16 November 2026, docket EPA-HQ-OAR-2025-0124).

So the rulebook that survives is chiefly the efficiency requirements for new gas turbines — provisions that matter to anyone building generation to serve data centres or a re-shored steel sector, but which do not impose CCS. AISI's language is accurate as a description of the CCS-dependent provisions; it overstates the scope as a description of the whole rule.

There is a second track. EPA issued a concurrent supplemental proposal that would rescind the 2015 greenhouse gas findings and repeal all Section 111 power-plant GHG standards on the basis that the Clean Air Act does not authorise such regulation at all. That is a proposal, not a final action, and EPA has said it is not reopening the final partial repeal. It is the mechanism by which the agency hopes to make the rollback durable against a change of administration. It is also the mechanism by which the litigation risk gets concentrated.

The legal footing: BSER, not endangerment

The repeal does not rest on a finding that power-plant CO2 is harmless. It rests on a re-evaluation of the best system of emission reduction (BSER) under Clean Air Act §111. EPA concluded that 90% CCS at existing coal steam units "has not been adequately demonstrated," that its costs "are not reasonable," and that the achievable degree of emission limitation was therefore unattainable. Crucially, the agency reversed a 2024 judgment: earlier projects that failed to reach 90% capture are now read as evidence the technology is not adequately demonstrated, rather than as evidence it is.

EPA also found that the capture, pipeline and sequestration infrastructure needed to apply 90% CCS across the existing coal fleet "does not currently exist and would need to be broadly deployed," and that this was "significantly unlikely" by the 1 January 2032 compliance date. On natural gas co-firing at existing medium-term coal units, EPA invoked West Virginia v. EPA (597 U.S. 697) to conclude that a 40% co-firing mandate would be impermissible "generation shifting," and separately found co-firing in a steam unit an inefficient use of gas under the §111(a)(1) energy-requirements factor.

The policy scaffolding cited in the rule is explicit: executive orders on "Unleashing American Energy," on coal, and on the asserted link between coal-fired generation and electricity demand growth from manufacturing re-shoring and AI data centres.

Why a steel association is arguing about power plants

This is the analytically important part, and it is broader than the boilerplate about electricity costs.

Steelmaking's electricity intensity is rising, not falling. Electric arc furnace (EAF) capacity — now the majority of US steel production — is a large and inelastic electricity consumer, and it is a demand problem as much as a price problem: an EAF cannot be curtailed without interrupting a heat. Integrated producers, meanwhile, face rising electrical load from pollution-control equipment, rolling and finishing lines, and the hydrogen and electrification pathways that are central to their decarbonisation plans. AISI's membership spans both, plus pipe and tube makers and processors.

The industry's exposure therefore runs through three channels:

1. Price. Wholesale power and capacity costs feed directly into EAF conversion costs, where electricity is typically one of the larger controllable items after scrap.
2. Firmness. Interruptible tariffs and demand-response obligations become more problematic as melting campaigns lengthen and as new continuous processes are electrified.
3. Availability for new load. Any capacity expansion — a new EAF shop, a DRI plant with electric process heating, a hydrogen-ready reheating furnace — depends on a grid operator's ability to serve it, which in several US markets is now the binding constraint rather than the capital cost of the equipment.

That is the context in which AISI's CCS argument becomes more than a generic anti-regulation posture. If the 2024 framework had stood, a meaningful tranche of the existing coal fleet would have faced a choice between retiring earlier than otherwise and undertaking CCS retrofits whose infrastructure did not exist. In the tightest capacity markets, that translates into higher capacity prices and a greater reliance on new gas — with the associated gas-turbine efficiency standards surviving. AISI's phrasing about "reliable and affordable" electricity should be read against that capacity arithmetic rather than as a generalised complaint about compliance costs.

The claim that repeal "will enable the American steel industry to continue to lead the world in low emissions and energy efficient steel production" is the weakest link in the statement. It is a political assertion, not an analytical one. US steel's emissions profile is driven overwhelmingly by the blast furnace/basic oxygen furnace versus EAF split and by the carbon intensity of the grid those furnaces draw from. Keeping older, less efficient coal units running raises the emissions intensity of the electricity that EAFs consume; on a Scope 2 basis, that cuts against, rather than supports, the leadership claim. The honest version of AISI's case is that grid reliability and cost come first, and that decarbonisation will be delivered through the steelmaking route and through efficiency — which is precisely what the statement says, but in a sequence that deserves to be made explicit.

The durability problem AISI does not address

AISI frames the repeal as settled: the action "will ensure that American steel producers continue to have access to reliable and affordable electric power." That is a stronger claim than the procedural position supports.

• The repeal is a single rule within a multi-part rulemaking. The supplemental proposal would go further and put the underlying statutory authority itself in play.
• The 2024 standards faced litigation from multiple states and regulated entities. Any final rule on this scale attracts challenges, and the West Virginia precedent cuts in more than one direction: it constrains EPA's use of generation-shifting BSER, but it does not resolve whether §111 supports the standards at all.
• The 2015 predecessor framework was itself partly vacated. Regulatory risk has not been eliminated by this rule; it has been redistributed among the supplemental rulemaking, the courts and the next administration.

For capital allocation, that matters. A steel producer weighing a decade-long electrification or capacity investment needs a view on power prices and firmness that survives an administration change — and a partial repeal plus a pending rescission proposal is a less stable foundation than the statement implies.

What to watch

• The effective date, 16 November 2026, and whether any party seeks a stay of the partial repeal pending review.
• Finalisation of the supplemental proposal. If EPA rescinds the 2015 findings, the legal ground shifts from BSER to statutory authority — a higher-stakes fight with a longer judicial tail. If it is withdrawn or narrowed, the partial repeal stands on its own BSER footing, which is more defensible but far less durable.
• Whether the retained turbine efficiency standards are revisited, given that new gas build is the most likely near-term response to load growth from data centres and re-shoring.
• Capacity-market outcomes in tight markets (notably PJM) as a read on whether the repeal actually delivers the affordability AISI cites. The rulebook change is upstream of the price; the auction results are where steel buyers will see it.

Assessment

AISI's endorsement is technically well-founded on the specific question it addresses — 90% CCS at the existing coal fleet by 2032 was not a credible BSER, and EPA's infrastructure finding is the strongest part of the final rule. The statement is less reliable on scope (the repeal is partial), on durability (a broader rescission is still pending and litigation is likely), and on the emissions-leadership claim, which confuses a power-sector outcome with a steel-sector one.

The real story for metals professionals is not the rule but the demand side behind it: US electricity load growth from data centres and industrial re-shoring is colliding with a generation fleet whose retirement schedule was, until now, partly set by environmental compliance. AISI has correctly identified reliability and price as the binding constraints for electric steelmaking. Whether this rule improves either, and for how long, is a question the Federal Register cannot answer — the capacity auctions and the courts will.