Data Centres Took a Third of Climate Tech's Capital

Climate tech companies raised $26.1 billion in venture capital in the first half of 2026, 55 per cent more than in the same period a year earlier and the strongest first half since 2022, according to the H1 2026 Climate Tech Investment and Innovation Report published by the finance tracker Currence on 21 September. Low-carbon data centre developers took 34 per cent of it.

Built Environment, the Currence category that holds those developers, rose more than 800 per cent year on year and displaced Energy as the largest vertical in the dataset. Grid technology had its best first half on record. The report landed during Climate Week in New York, where MIT Technology Review reported on 24 September that AI had taken over the agenda, and against the backdrop of a UN Environment Programme assessment published on 2 September concluding that warming will exceed 1.5C and that the remaining question is how briefly.

Two deals, a quarter of the money

The growth is concentrated to an unusual degree. Currence counted a 25 per cent fall in deal count, a five-year low, while the ten largest deals absorbed 42 per cent of all funding and Series C rounds took 40 per cent of the total, up from 16 per cent a year earlier. Two transactions did much of the work: DayOne's $4.5 billion Series C, which Currence calls the largest climate tech deal since 2020, and Nscale's $2 billion Series C. Both are data centre developers, and together they account for roughly a quarter of the half.

That is a market with more money in it and fewer companies receiving any. The distinction matters because concentration of this kind gets read as sector health when it is closer to the opposite. A half in which ten cheques carry 42 per cent of the total, and in which the median company finds it harder to raise than it did a year ago, is not a sector-wide recovery, even though the aggregate line moves the same way a recovery would.

What stopped getting funded

The same report shows where capital left. Currence describes fuels as still on pause and says carbon equity funding collapsed. Transportation rose 52 per cent. Climate Management, the monitoring and risk category, had its best first half since 2022, up 6 per cent, which Currence attributes to rising physical climate risk and cheaper satellite capacity pulling money toward measurement.

The pattern is legible. Technologies a data centre can buy are being financed. Technologies that only a carbon market or a fuel mandate can pay for are not. Clean firm power, grid equipment, cooling and power optimisation now have a creditworthy customer with a construction schedule attached. Direct air capture and low-carbon fuels have policy instead, and policy in the United States has moved against them.

Adaptation turns out to be investable

The largest deal outside energy and transport was ICEYE's $521 million for radar satellites. Currence frames that as adaptation arriving as a growth-stage asset class rather than a grant category, on the argument that physical climate risk has stopped being a forecast and become an observable, insurable loss. It is a real shift, and it is happening for the same reason the data centre deals are happening: somebody now has a bill they need to price.

The honest reading is that this is good for deployment and bad for the pipeline. Capital chasing an offtaker is how a technology crosses from pilot to commercial scale, and clean firm power is getting exactly that treatment, with a record half for clean power listings led by Fervo Energy and X-energy. As Parallax Nexus reported on 14 September, Fervo's 396 MW geothermal deal with Google was already showing how hyperscaler procurement decides which clean-firm projects get built. But the sectors being starved are the ones with the longest development timelines and the least near-term revenue, which is precisely the profile of what the 2030s will need.

The counter-argument is straightforward and partly right: a dollar into grid technology or geothermal is a decarbonisation dollar whatever motivated it, and the emissions arithmetic does not care who the buyer was. What is not known is how durable that demand is. Currence's own figures describe a half in which two deals set the tone, and a market that concentrated can turn just as quickly if data centre construction forecasts are revised down.