Brookfield and Bloom Energy Expand AI Power Partnership to $25 Billion

Brookfield and Bloom Energy have expanded their partnership to finance power for artificial intelligence infrastructure, lifting the framework from $5 billion to $25 billion in a fivefold increase since October 2025. The enlarged commitment will fund the global deployment of Bloom's fuel cell systems to supply fast, onsite electricity for AI data centres, an area where demand has outrun the grid's ability to connect new load. The move deepens a collaboration that pairs Bloom's rapidly deployable power platform with Brookfield's capital and infrastructure scale, and it strengthens the investment firm's positioning as a leading backer of AI infrastructure.

 

Why AI Needs Power the Grid Can't Supply Fast Enough

 

The rationale rests on a mismatch that has become the defining constraint on AI expansion. Hyperscalers and data centre developers need large volumes of reliable power quickly, but connecting new demand to the grid can take years because of interconnection queues and transmission bottlenecks. That gap between how fast AI infrastructure is being built and how slowly grid power arrives is precisely the space Bloom's technology is designed to fill.

Bloom's answer is what the companies call islanded, or onsite, power. Its fuel cell systems generate electricity at the data centre itself rather than drawing it from the grid, which lets a developer power a facility without waiting for a grid connection. The systems are marketed as clean, reliable and quick to deploy, and the appeal to AI operators is speed above all: a power source that can be installed on the timescale of a data centre build rather than the timescale of a transmission upgrade.

 

Read more: The Protein Brewery Closes €18 Million Series B Extension Led by ABN AMRO

 

An Integrated Model From Electrons to Compute

 

The partnership is structured around more than just supplying hardware. The two companies are advancing what they describe as a new model for AI factories, one that integrates power, compute, data centre infrastructure and capital from the outset rather than assembling them piecemeal. Brookfield brings the capital and development scale, Bloom brings the onsite generation, and the combination is pitched as an end-to-end offering.

That integration is the strategic heart of the deal. Brookfield's AI infrastructure head, Sikander Rashid, framed the firm's ambition as delivering solutions spanning "from electrons to tokens," capturing the idea that AI infrastructure investors increasingly want to own the whole stack, from the power source through to the compute that runs on it. By locking in a dedicated power partner at scale, Brookfield positions itself to offer sophisticated customers a fully integrated package rather than competing solely on capital, and Bloom secures a financed route to deploy its systems on the large projects driving demand.

 

Explore OneStop ESG Marketplace: Renewable Energy

 

A Piece of a Much Larger AI Bet

 

The expanded framework sits inside a far bigger Brookfield strategy. The commitment forms part of the firm's dedicated AI Infrastructure Fund, launched in November 2025 with a target to deploy $100 billion across AI factories, power, compute and strategic partnerships, and Brookfield says it has already invested more than $100 billion in digital infrastructure and clean power assets. Against that backdrop, the $25 billion Bloom framework is one building block in a portfolio-level push rather than a standalone position.

For Bloom, the scale-up is a significant commercial validation, giving it committed financing to grow its fuel cell business globally at a moment when its islanded power solutions are in rising demand. The larger question the deal raises is how much of the AI power problem gets solved onsite rather than on the grid, and whether fuel-cell generation at this scale proves as clean, reliable and cost-effective as promised once deployed across large projects. As with any framework of this size, the figure signals intent rather than committed spend, so the pace at which capital is actually deployed, and the performance of the systems as they come online, will show whether this model becomes a standard route to powering AI or one option among several.

 

Source: Bloom Energy