§ 01 · Twenty years of carbon price

How we got here, in numbers.

The EU ETS launched in 2005. The European Commission’s reference page on the system sets out the underlying legislative framework and the four trading phases the market has now passed through. The price history matters because it shapes how most operators think about the carbon column in their budgets — and most of those mental models are out of date.

Phase 1 (2005–2007) was the learning-by-doing phase. Allowances were issued on the basis of estimated emissions, the supply substantially exceeded actual emissions, and by 2007 the EUA price had collapsed to effectively zero because phase 1 allowances could not be banked into phase 2. Industrial operators who paid any attention at all to the carbon market during this period learned that it was not a serious cost driver.

Phase 2 (2008–2012) coincided with the first commitment period of the Kyoto Protocol. The cap was tighter and based on actual emissions data, but the 2008 financial crisis produced emissions reductions that exceeded the cap reductions, creating a large surplus that pushed prices into the €5 to €15 range for most of the phase. The lesson reinforced for operators: the carbon market reacts to macroeconomic shocks, surpluses persist, and the underlying price signal is weak.

Phase 3 (2013–2020) introduced an EU-wide cap, auctioning as the default allocation method, and broader sector coverage. Prices spent most of the phase between €5 and €30, finally rising into the €25 to €30 range from 2018 onward as the Market Stability Reserve started withdrawing surplus allowances. The first sustained period in which carbon began to register as a real cost for some heavy-emitting installations.

Phase 4 (2021–2030) is the current trading phase. The tighter cap from the Fit for 55 package, the planned 62 percent emissions reduction by 2030 versus 2005 levels, the gradual elimination of free allowances, and the inclusion of maritime transport from 2024 have all pushed the supply-demand balance toward structural deficit. Prices crossed €60 per tonne in early 2022, briefly reached €100 per tonne in February 2023, retreated to the €60 range during 2024 on weak industrial demand, and crossed back above €80 in Q4 2025 as traders began positioning for the 2026 supply shortfall.

The picture this puts together is straightforward. The carbon price was structurally low for fifteen years, became economically meaningful around 2021, and is now firmly in territory where it changes operational decisions for in-scope industrial operators. The trajectory points up from here, with the supply-side dynamics in 2026 and 2027 being the most acute in the system’s history.

§ 02 · The 2026 supply deficit

Four drivers behind the structural shift.

The tightening emissions cap. The fourth-phase cap declines at a Linear Reduction Factor of 4.2 percent per year, accelerated from earlier phases. By 2026 this translates to roughly 110 million tonnes fewer allowances issued than in 2024, before any other adjustments are applied. The cap reduction was decided in 2023 as part of the Fit for 55 package and is now baked into the system; the supply trajectory is mechanical from here through 2030.

The phase-out of free allocations. Industrial installations have historically received a significant share of their allowances for free, calibrated to product-specific benchmarks. Free allocations are now being scaled down progressively, conditional on each installation’s decarbonization progress. The aviation sector loses free allocations entirely in 2026. For industrial operators, the practical effect is that the share of total emissions covered by free allocations is shrinking each year, and the share that must be purchased at market price is correspondingly growing.

The inclusion of maritime transport. Shipping emissions entered the EU ETS scope in 2024, with a phase-in to full coverage by 2026. This adds substantial new demand for allowances — on the order of 70 to 90 million tonnes of CO2 annually at full coverage — without a corresponding increase in the cap. The new demand is structural rather than cyclical.

Market Stability Reserve dynamics. The MSR removes allowances from circulation when the total number of allowances in circulation exceeds defined thresholds, and injects them back when scarcity emerges. The European Commission proposed in April 2026 to stop the invalidation mechanism that had previously cancelled allowances above 400 million in the reserve, allowing the buffer to grow. This proposal — if adopted — preserves more flexibility for the Commission to intervene if prices rise too aggressively, but does not materially change the underlying 2026 supply position.

Taken together, major bank carbon desks estimate a 2026 supply reduction of approximately 180 million tonnes year-on-year. Demand projections vary — industrial demand remains soft, power-sector emissions are below historical averages thanks to renewables expansion and gas-to-coal fuel switching — but even with weak demand, the supply contraction is sufficient to drive prices upward. The consensus end-2026 price forecast across major desks sits near €100 per tonne, with downside risk if European industrial demand stays weak or a Russia–Ukraine peace agreement materially reduces gas prices, and upside risk if the supply-side dynamics tighten further than currently projected.

§ 03 · The operational implications

What the carbon column actually means now.

For an industrial operator in 2026, the carbon price meaningfully changes four operational considerations.

Fuel switching economics have inverted. The relationship between coal and gas in European power generation has flipped twice in the last three years. At €30 per tonne CO2, coal was cheaper than gas for most generation profiles. At €80 per tonne CO2 with current gas prices, gas is now meaningfully cheaper than coal on a fully-loaded cost-per-MWh basis, even after accounting for capacity factor differences. For industrial operators with internal cogeneration assets or significant indirect exposure through power purchase agreements, the carbon-adjusted economics of the fuel mix have to be re-evaluated at least annually rather than treated as a fixed input.

Capex prioritisation has shifted. Energy-efficiency projects that previously had marginal NPV at €30 per tonne CO2 frequently clear the hurdle rate at €80 to €100 per tonne. A heat-recovery project saving 5,000 tonnes of CO2 per year was worth approximately €150,000 annually at 2022 prices; at consensus 2026 prices it is worth approximately €500,000 annually. Projects that were deferred for five years on payback grounds in 2020 may now have 18-month paybacks. Operators that have not refreshed their internal carbon-cost assumptions in the last 24 months are systematically under-investing in decarbonization opportunities that genuinely make economic sense.

Procurement hedging is no longer optional for large emitters. An installation emitting 100,000 tonnes of CO2 annually faces a €10 million carbon line item at €100 per tonne. Treating that exposure as floating-rate is the same financial decision as treating €10 million of gas exposure as floating-rate, which most large industrial operators stopped doing decades ago. Forward purchasing of allowances, financial hedging via futures, and structured contracts with carbon-cost pass-through clauses to customers are all now standard practice and should be part of any large emitter’s annual financial planning rather than an ad-hoc activity by the sustainability team.

The downstream pass-through question is sharper than before. For some industrial products — cement, primary steel, ammonia — carbon cost is now a meaningful share of price. Whether the operator can pass that cost through to customers is determined by market structure, import competition, and CBAM coverage. CBAM theoretically equalises carbon cost between EU and non-EU producers for covered products from 2026, but the practical effectiveness of CBAM for the first two years remains contested, and operators in CBAM-covered sectors need to model multiple scenarios for pass-through behaviour rather than assuming one outcome.

§ 04 · The honest downside risks

Three scenarios in which the price forecast does not play out.

Russia–Ukraine settlement. A negotiated peace agreement that restores meaningful natural gas flows from Russia to Europe would drive European gas prices materially lower. Because gas is the price-setting fuel for marginal European power generation, lower gas prices reduce the value of switching from gas to lower-carbon alternatives, which reduces demand for emissions allowances. ABN AMRO and other bank desks specifically flagged this as the largest single downside risk to their 2026 EUA price forecasts.

European industrial recession. European manufacturing PMI remained below 50 for most of 2025, indicating contracting industrial activity. A deeper recession in 2026 — driven by US tariffs, weak global demand, or further deindustrialisation pressure — would reduce industrial emissions and demand for allowances. The supply-side contraction would still drive prices upward, but more slowly and to lower levels than the consensus forecast.

Commission intervention. The European Commission has tools to manage the carbon price if it rises too quickly. The Article 29a price-stability mechanism allows additional allowances to be released if the price exceeds certain thresholds. The Commission’s April 2026 MSR amendment was framed as enhancing stability and predictability, and a full comprehensive review of the EU ETS is scheduled for July 2026. If the Commission concludes that prices are rising too aggressively for industrial competitiveness, intervention to release allowances or accelerate front-loading is possible. This would not eliminate the upward trend but would limit the rate of increase.

None of these scenarios is sufficiently likely to invalidate the underlying upward trajectory, but each is sufficiently probable that operators planning around a hard 2026 price forecast should model a range rather than a point estimate. A reasonable planning range for end-2026 EUA price is €70 to €110 per tonne, with the central estimate sitting around €95 to €100.

§ 05 · Bottom line

Three things operators should actually do.

Update your internal carbon cost assumption. If your capital-allocation models still use a €40 or €50 per tonne carbon cost for project NPV calculations, refresh that assumption to a planning range of €80 to €110 for projects evaluated in 2026 and beyond. Many decarbonization projects that did not clear the hurdle rate two years ago will clear it now. The opportunity cost of running on outdated assumptions is real and growing.

Build a coherent allowance position rather than buying as needed. For installations emitting more than 25,000 tonnes per year, ad-hoc allowance purchasing exposes the operation to short-term price volatility and removes a meaningful tool for budget predictability. Forward purchasing some share of expected annual emissions, supplemented by futures or options for the balance, is now standard practice for serious emitters and should be on the agenda for any operator that has not yet adopted it.

Plan for the trajectory rather than the current price. The 2026 supply deficit is the start of a multi-year tightening, not a one-off event. Operators planning capex programs that will be commissioned in 2028 or 2029 should model carbon costs at the trajectory level, not the current level. The carbon column on the operating budget is going to be a structural part of industrial economics for the rest of this decade and beyond. Treating it that way starting now is materially cheaper than catching up to it later.