The World Bought Less Clean Power in 2025. Japan's Corporate Buyers Kept Signing.

BloombergNEF’s count for 2025 is the first decline in the corporate clean-power market in nearly a decade. Companies announced deals for 55.9 GW of clean power over the year, down 10% from the record set in 2024 (BloombergNEF, published 19 February 2026).

The regional split in that number is worth reading slowly. The Americas took 29.5 GW. Europe, the Middle East and Africa took 17 GW. Asia Pacific — every market from India to Australia to Japan and Korea — took 6.9 GW.

Two other figures from the same count explain the decline better than the headline does. Meta, Amazon, Google and Microsoft together accounted for 49% of global activity, with Meta and Amazon alone signing 20.4 GW. And the number of unique corporate buyers in the United States fell to 33, down 51% year on year.

So the market did not shrink because companies stopped buying clean power. It shrank because the set of companies doing the buying got smaller, and what remains is increasingly four balance sheets procuring for data centres.

Japan’s market has the opposite shape

Japan Energy Hub, which maintains a database of disclosed Japanese corporate power purchase agreements, puts the cumulative count past 500 deals since the first corporate PPA in Japan was signed in 2021, representing more than 2.5 GW of installed and planned capacity (Japan Energy Hub).

Five hundred deals for two and a half gigawatts averages roughly five megawatts a deal. That is one array behind one factory, not a hyperscaler contracting a wind farm.

The two numbers do not divide into each other. BloombergNEF counts announced deal volume in a single year; Japan Energy Hub counts disclosed deals cumulatively since 2021, installed and planned. Different metrics, different clocks. Together they support a claim about shape rather than size: the global market is concentrating into a few very large buyers, and Japan’s is not.

A market where four companies are half the demand moves when four capital-expenditure committees move. A market of five-megawatt deals across hundreds of buyers has no single point of failure, and no single point of acceleration either.

What the supply side is being asked to do

Japan’s seventh Strategic Energy Plan, published in February 2025, sets the target the corporate buyer is implicitly financing. Renewables go from 22.9% of the power mix in FY2023 to 40–50% in 2040. Nuclear goes from 8.5% to about 20%. Thermal generation falls from nearly 69% to about 30–40% (Enerdata).

Those shares sit under a national emissions target of a 60% cut by FY2035 and 73% by FY2040 against FY2013 levels, taking annual emissions from 1.4 GtCO₂e to 570 MtCO₂e and then 380 MtCO₂e (same source).

A target does not build anything. Someone has to sign a contract long enough for a lender to underwrite the asset, and in a market where merchant revenue will not carry construction debt, the corporate offtaker is one of the few counterparties who can. That is the actual function of the corporate PPA in Japan, and it is the reason the deal count is a better indicator of the transition’s health than the megawatts.

The carbon price is not what is driving it

The common explanation is that Japanese companies are buying clean power because carbon is about to cost money. That explanation does not survive contact with the schedule.

The GX-ETS entered its mandatory phase on 1 April 2026. It covers entities emitting 100,000 tonnes of CO₂ a year or more, and for those entities surrendering allowances became a legal obligation rather than a pledge (ICAP). We wrote up the mechanics when it took effect, in Japan’s GX-ETS went mandatory.

The part that matters here is what a covered company pays. Allowances are allocated, not sold. Auctioning does not begin until FY2033, and even then only for high-emitting corporations in the power sector.

An allowance you receive for free is a reporting obligation, not a price signal. Seven fiscal years separate the compliance obligation from the first auction, which is longer than the tenor of many of the PPAs being signed now. Whatever is pulling Japanese corporates into these contracts, it is not the cost of a domestic allowance, because for most of them that cost is currently zero.

Our reading of what is actually pulling

The pressure is arriving through the customer, not the regulator. We have no survey to cite for that, so take it as our reading. A Japanese manufacturer selling into Europe is being asked for Scope 2 evidence by a buyer with its own disclosure obligation, and the evidence has to be specific — this site, this year, this contract — rather than a corporate-level renewable claim. A five-megawatt onsite array behind one plant answers that question for that plant in a way a national certificate purchase does not.

That is consistent with the deal size. If the driver were a carbon price, you would expect large corporate-level procurement optimising an allowance position. If the driver is a customer asking about a specific site, you get exactly what Japan has: hundreds of small, site-attached contracts.

It is also consistent with the timing. The obligations pulling on Japanese suppliers from the European side are landing before FY2033, not after.

Three things worth watching

Whether virtual PPAs become routine. A physical contract tied to one site cannot serve a company with fifty of them. The financial structure is what lets procurement scale past the roof it sits on, and it is the structure with the most unsettled accounting and regulatory treatment in Japan.

Whether certificates undercut contracts. Non-fossil certificates are the cheap way to answer a Scope 2 question. If buyers conclude that a certificate satisfies their European customer, the additional capacity a PPA finances does not get built, and the deal count keeps rising while the megawatts do not.

Whether the grid can absorb what gets signed. We looked at this from the supply side in Japan’s battery storage queue: applications to connect standalone storage reached 170.8 GW against a small fraction holding a connection contract. A signed PPA and an energised asset are separated by the same interconnection process, and the second number is the one that decarbonises anything.

The useful indicator, for anyone modelling this market, is not the annual gigawatt total. It is the number of distinct corporate buyers signing in Japan each year. If that number climbs while the global count of unique buyers falls, Japan is running a genuinely different market — broad rather than concentrated — and it will be more durable through a downturn than the one BloombergNEF just measured.

Tech for Impact Summit 2027 takes place in Tokyo on 18–19 May 2027, co-hosted by the United Nations University. It is invitation-only, and questions like this one — where a policy target stops and a corporate signature has to start — are the kind the room is built around. If you are working on them, you can ask for an invitation.