Working back from 2031

Opinion Piece · October 2026
By Clément Georget, Co-Founder, Rainbow

The Commission has given permanent carbon removals a start date in the ETS: 2031. Engineers read start dates backwards.

A few weeks ago, a developer asked me a question I now hear in most calls. He runs a biomass plant and is deciding what to build next to it. He could add capture to it. The question was if anyone would buy the tonnes, if they’d buy all of them, and from when they would buy them. If the answer is not clear, the CO₂ keeps going up the stack.

I did not have a clean answer, but I had a date. The Commission has proposed that, starting in 2031, the ETS will pay for permanent removals. They will auction 250 million extra allowances over ten years and use the money to buy certified BioCCS and DACCS tonnes. This is the first time Europe’s carbon market has named a year for removals. It is also, if you read it the way the developer does, already late.

This piece is split in two parts. The first sizes the problem the way an engineer sizes a system: load, installed capacity, lead time. The second counts down from 2031 to today and asks, year by year, what has to be true already.

Part 1: The sizing

The load. The load is how much a system has to deliver. Here, we mean the permanent removals Europe still needs after its cuts: 27 to 75 million tonnes a year by 2040. The Commission’s own 2040 impact assessment has gross emissions falling to around 850 million tonnes a year, three quarters below 1990. Even then, reaching the 90% net target needs up to 390 million tonnes of removals. Most of that is forests and soils. The rest is industrial: BECCS and DACCS. By 2050 the same scenarios lean on close to 180 million tonnes a year of permanent removals. That is roughly the annual emissions of the Netherlands, removed every year, forever.

The installed capacity. Installed capacity is how much the system can deliver today. Worldwide, that is hundreds of thousands of tonnes a year. Europe hosts almost half of the world’s permanent removal projects, and most of them are not BECCS or DACCS plants. They are biochar sites, pyrolysis units bolted onto farms and sawmills, biomass boilers with a capture retrofit on the drawing board.

The ETS proposal’s 250 million tonnes is a ramp: a few million tonnes bought in 2031, rising to around 48 million a year by 2040. The smallest purchases fall in the years when plants most need a buyer, and the 2040 endpoint is still about a quarter of what the same scenarios need every year by 2050.

The lead time. Lead time is how long it takes to build what is missing. The gap is a factor of fifty to a hundred in fourteen years. No industry has grown at that rate without demand arriving years before the capacity was needed. Wind and solar had feed-in tariffs a decade before they mattered on the grid. Offshore wind had contracts for difference before the first monopile was ordered. The demand instrument always comes before the steel; that is not a policy preference, it is how lenders work.

The obvious objection is that the land sink does most of the 2040 arithmetic anyway. It does, on paper. It has also been shrinking for a decade and can reverse in a single fire season. A fossil tonne stays in the atmosphere for centuries; it cannot be balanced by a sink that may not be there in thirty years. The ETS proposal itself accepts this: it matches fossil emissions only with permanent units. Good. Then the permanent units have to exist.

The result. Put the load, the capacity and the lead time together and they reduce to one sentence. If the first ETS purchase is in 2031 and the 2040 need is around 50 million tonnes a year, nine years of purchases have to carry the whole scale-up. This means the plants delivering in 2031 must be financed in 2027, at the latest.

Part 2: The schedule

2031 - First purchase. From 2031, the Commission auctions up to 250 million extra allowances and uses the revenue to buy removal units certified under the CRCF, limited for now to BioCCS and DACCS. The Commission is the only buyer; it purchases the units and cancels them. By the end of 2034 it reports on whether operators might buy removals directly, and whether nature-based units might join.

2030 - Verified removal. A unit purchased in 2031 must be verified in 2030, on tonnes captured in 2030.

2029 - Commissioning. For a plant to be running in 2029, somebody signed the contracts that let it be built. Not a target in a directive: a counterparty, a price, a delivery window and a definition of the unit that will be paid for. That is what a demand signal is to a lender. Anything less is a press release.

Look at Stockholm Exergi, the only European BECCS plant of real size that has actually reached a final investment decision. It ran a pilot from 2019. It secured a €180 million Innovation Fund grant, then a fifteen-year contract from the Swedish state worth SEK 20 billion, then a ten-year offtake with Microsoft for 3.33 million tonnes. Only then, in March 2025, did its board take the decision to build. Its first tonne will be delivered in late 2028. That’s six years from pilot to decision, and three and a half years from decision to first tonne. Three demand instruments were stacked before a lender would sign, and none of them was the ETS.

That is why the industry’s ask for long-term contracts by 2029 is not an early date. For a plant delivering in 2031, it is already late.

2028 - Design freeze. A year before commissioning, the design is frozen. The engineers need one answer from Brussels before they freeze it: will this tonne count?

The proposal says only two technologies count: BioCCS and DACCS. Everything else waits for a report in 2034.

I understand the instinct: start with what is unambiguously permanent. But a list of methods is the wrong tool for the job, and we already have the right one. The CRCF sorts every unit by what happens to the carbon: permanent storage, storage in products, carbon farming. It already runs biochar on a 200-year basis. It will sort mineralisation, ocean alkalinity enhancement and enhanced weathering the same way as they come through. A durability basis answers the 2028 question for every plant on the drawing board.

Remember what Europe’s project base looks like: it has almost half of the world’s permanent removal projects, and most of them are not BECCS or DACCS. If the design input in 2028 is “your tonne does not count“, those plants get sited for other buyers, and Europe’s first-mover advantage becomes a line in a 2034 report about a market that moved.

None of this loosens the like-for-like rule. A permanent fossil emission is matched by a permanent unit. The question is only who sorts the units: a list written in 2026, or a certification framework built to do exactly that.

2027 - Financing. Two years before commissioning, the money is raised. The lender asks who else is buying, and the honest answer is: one public buyer, from 2031, for two technologies. A single buyer is a floor. It is not a market, and a lender prices the difference.

And 2027 is the generous version. Stockholm Exergi took three and a half years from decision to first tonne. On that timeline, a plant capturing in 2030 takes its investment decision in 2026 - this year.

The voluntary market already taught us this. A handful of buyers set the rules for everyone, and when one of them changed its methodology preferences, whole pipelines turned. Replacing a few private buyers with one public one is not diversification. The same developer taking a decision in 2027 is weighing the ETS against a Microsoft or Frontier offtake, a national reverse auction in Sweden or Denmark, a US project with tax credits. The schedule is identical everywhere. The only question is which counterparty makes it bankable first.

Three things change the lender’s view in 2027. Letting operators buy eligible units directly, within a quota, so the ETS has more than one buyer from day one. Letting intermediaries aggregate small projects into contracts a lender can read. And closing the price gap. An allowance trades at a fraction of what a permanent removal costs, so auctioning 250 million allowances will not pay for 250 million tonnes; the industry puts the shortfall at €16 to 20 billion over the decade. The proposal already creates the instrument that could cover it: the Industrial Decarbonisation Bank, which starts lending and granting in 2028, three years before the first removal auction. It only works if permanent removals are eligible from its first call, not added later.

Today - The plumbing. This brings the countdown to October 2026, and to the part of the system nobody writes position papers about.

For a tonne to be bought in 2031 it has to travel: from a methodology, through a certification scheme the Commission has recognised, past a verifier with the capacity to audit it, into a registry, and from that registry into the Union Registry where allowances live. Each link is being built now. Scheme recognition under the CRCF takes quarters, with accreditation to match. Verifiers are training up on removal methodologies. Registry-to-Union-Registry integration is being tested. None of it is out of reach by 2031. The ETS proposal does point to the CRCF: only units certified under it qualify. But pointing at a framework is not the same as scheduling it.

This is why both science and field engineering matter. A methodology defines what a removed tonne is, and that work belongs with scientists. But a tonne is not removed by a methodology. It is removed by a project that does the work on the ground: equipment that runs as designed, measurements taken in the field, and records that hold up when someone checks them. That is field engineering. Even then, the tonne only counts once the rest of the chain is in place: someone financed the plant, a verifier checked it, a registry issued the credit, and a buyer paid for it.

Back to the developer

I called the developer back. I told him the ETS would pay for permanent removals from 2031, that his tonne probably would not count under the current text, and that nobody could yet tell him who else would be willing to buy it. He said he would wait a year and see.

The Commission has set a start date of 2031. Developers in Europe are counting back from it, and what they see is that their plants need to be financed years earlier, long before the rules give them a reason to. Many are now waiting for one sentence in the proposal to be rewritten in 2034: the one that limits which removals the ETS will buy.