There is a sentence buried in the newly released UN Environment Programme’s report on overshooting 1.5°C that reads less like science than confession. The pathway it recommends, the authors write, is “simply the best remaining option.”

For a decade the climate debate was organized around avoiding 1.5°C. UNEP now says that debate is over. Warming will cross the line within a few years. Even if every national plan and net-zero pledge were fully delivered, an optimistic assumption, the world would peak near 1.8°C. The goal hasn’t changed, but we will have to approach it from above.

Cutting emissions to zero only stops the warming. To bring temperatures back down, you must remove carbon dioxide already in the atmosphere. The report’s best estimate is that shaving one tenth of a degree off global temperatures takes about 220 billion tonnes of net-negative CO2. Humanity emits more than 40 billion tonnes a year. Planting trees at today’s global rate would take roughly a century to deliver that single tenth.

So the report is blunt about two things at once. Carbon removal cannot substitute for cuts; even optimistic scaling yields a few tenths of a degree this century, but removal is nonetheless mandatory. Every pathway back to 1.5°C leans on it. Engineered approaches such as direct air capture supply less than 0.1 percent of removals today and need to reach somewhere between 0.9 and 6.4 billion tonnes a year by mid-century.

That collides with a second, quieter number. Models suggest ten billion tonnes or more of residual CO2 will still be emitted annually at net zero, from cement, steel, aviation and chemicals. Every residual tonne must be matched by a removed tonne before removal contributes anything to cooling.

This is why capturing carbon at smokestacks and pulling it from the air are not competing priorities. They are the same project. Cement alone is roughly eight percent of global CO2 , and most of it comes from chemistry, not fuel: heating limestone releases CO2 no matter what powers the kiln. Heidelberg Materials’ Brevik plant in Norway, the first commercial-scale capture facility on a cement kiln, now captures about 0.4 million tonnes a year. That is real, but it’s also nearly alone. Global Cement’s May tally projects that under two percent of the industry’s emissions will be captured by 2035, and that flagship projects in Belgium and Sweden have stalled.

Why? The honest answer is that the world’s attention, and its capital, went elsewhere. The five largest U.S. cloud providers plan to spend between $660 and $690 billion on AI infrastructure this year. Britain’s Defence Investment Plan commits £298 billion over four years. The OECD’s June outlook warns that sustaining such budgets may push governments to postpone climate action. Money for hard, early technologies is thinning: seed and Series A climate investment fell in 2025 for the first time in about a decade, and early-stage investors’ share of climate funding dropped from roughly 20 percent in 2021 to under 8 percent. That is bad news for carbon capture, low-carbon cement and green steel specifically.

These priorities are not irrational. A fractured world does require defence. But notice the structure of the trade. Defence and AI are things a nation or a company can buy for itself and benefit from alone. Carbon removal is the opposite: a public good that pays off only if many countries build it, verify it, and share storage across borders. UNEP is explicit that credible removal depends on measurement standards and multilateral governance that do not yet exist at scale. We are funding the problems that fit our politics and starving the one that doesn’t.

To be fair, policy is moving, albeit slower than is needed. On July 17 the European Commission proposed folding permanent removals into its emissions trading system for the first time: a public authority would buy 250 million tonnes of direct air capture and bio-CCS removals between 2031 and 2040, financed by auctioning allowances. That is the first serious attempt to create compliance-grade demand for removal, rather than leaving it to corporate goodwill. In Washington, the same 2025 bill that gutted much of the Inflation Reduction Act left the 45Q credit intact at $85 a tonne for point-source capture and $180 for direct air capture. And in April, Ottawa widened its CCUS investment tax credit while Alberta committed to an industrial carbon price reaching $130 a tonne by 2040.

UNEP’s own framing is the one to end on. Overshoot is not an argument for resignation. It is an argument for urgency, and for building, now, the industry that will have to clean up after us.

Disclosure: Phil De Luna is CTO and co-founder of CURA, which develops electrochemical decarbonization technology for cement and lime, and was previously Chief Science and Commercial Officer at the direct air capture company Deep Sky.