DAC’s bumpy road to commercial scale

The concrete fact the episode turns on is a gap. Orca, Climeworks’ first plant, has a nominal capacity of 4,000 tons of carbon dioxide a year, meaning what the design would capture running at maximum efficiency. About two and a half years after starting up, it runs at roughly 2,500 to 3,000 tons. Nothing about the underlying chemistry failed. Capturing carbon at 420 parts per million, one molecule in 2,500, is genuinely hard, but that difficulty was known. What Climeworks did not know was everything that only appears when you run 24/7 in a real operating environment and owe a customer deliveries. A filter supplier’s quality varied within tolerances that were fine for that supplier’s other customers and not fine for capture efficiency measured over weeks and months. Iceland’s weather arrives as it is, with snow, hail, horizontal rain and impurities, and pre-treating incoming air is uneconomic almost immediately, which is why their chief technology officer at the time described the job as building a tractor rather than a sports car. Sequestration demands very pure carbon dioxide delivered continuously, so without a downstream purity upgrade you can end up dumping gas you already spent energy capturing. And the accounting compounds it: certification deducts the lifecycle emissions of building and operating the plant, and that deduction starts on day one, when the plant is least efficient, so the early months sell an even smaller share of what they capture.

The financing strategy is a direct consequence. Aepli argues you cannot go from pilot scale to the size at which a plant is finally profitable in one jump, because that concentrates enormous technological risk into a single very large capital commitment. Instead you take digestible steps of roughly 5 to 10x, which de-risks the technology at lower capital per lesson. Orca at 4,000 tons was followed in May by Mammoth at almost 10x, and the next plant, the first in the United States, is around 250,000 tons, another 7 to 8x. But every intermediate plant is by construction not yet profitable at its own size, which is precisely what project lenders will not fund. That is why the 2022 raise was a roughly $650 million corporate equity round rather than project debt. The eventual goal is a diversified stack of grants, project equity, project debt, tax equity and some corporate equity, and the pieces arrive gradually as proof accumulates. Climeworks did get a project finance facility done against Orca, which Aepli says was hard and had to be subsidized because Orca is not profitable at that size. For the next plant he expects allocated government grants plus project equity, with a classic project financing possibly only after the plant has proven operational stability through its ramp-up. One detail makes that more plausible than it sounds: Climeworks’ own technology is less than half the capital cost of a project, and the standard balance-of-plant equipment can carry a construction contractor’s performance wrap even when the novel part cannot.

Then the market description inverts the obvious reading. Today direct air capture looks supply-constrained, and in one sense it is: Orca and Mammoth are sold out for their expected operating lives and Climeworks is now selling the next plant. But that is not the constraint that decides whether the next plants exist. To justify a facility of 200,000 to 500,000 tons a year or larger you need offtake contracts running to millions of tons, and Aepli says he would want 10 to 15 million tons committed for a single million-ton facility. The entire market has sold 5.7 million tons so far this year. So the market is short of delivered tons and short of bankable long-dated demand at the same time, and by his account it is the second that limits how fast new plants can be committed. He expects that to flip, but he dates the flip carefully: not the 2024, 2025, 2027 or 2028 vintages, which are thin, but the 2030 and 2035 vintages, as companies come up against net-zero commitments they have to actually fulfill.

That is also why transparency functions as strategy rather than virtue. A lot of new direct air capture companies have appeared, funded in an environment where equity rewards ambitious promises, and Aepli’s fear is industry-level: if cost targets and deployment volumes are over-promised and then missed, buyers conclude the whole category promises and does not deliver. He would rather competitors deploy, hit the same learning curve, and inject what he calls a dose of reality, even though he expects some of them will not survive the step. The same logic runs through his contracting. There is no market price for removal, so pricing balances cost-plus against what buyers will pay, and Climeworks has spent about four years pushing customers toward terms that make assets financeable: five years minimum, ideally ten or fifteen, take-or-pay, clear payment terms, and assignable to the asset being financed. Customers push back, he says, because few sellers ask for any of it.