Bloom Energy and Brookfield Quintuple AI Power Deal to $25 Billion as Data Center Demand Surges

The race to power artificial intelligence just reached a new scale. On June 30, 2026, Bloom Energy and Brookfield Asset Management announced they were expanding their strategic AI infrastructure partnership from $5 billion to $25 billion—a fivefold increase that reflects the growing consensus that the electrical grid simply cannot keep up with what the AI industry needs.

Bloom Energy’s stock surged 12% on the announcement, a market reaction that captures both investor enthusiasm and an underlying acknowledgment: the power bottleneck has become the defining constraint on the AI infrastructure buildout, and any credible solution commands a premium.

The Power Problem at AI’s Core

The numbers behind AI’s energy appetite are difficult to overstate. Worldwide data center power demand is projected to rise 27% in 2026 alone, reaching 132 gigawatts of total installed capacity. Goldman Sachs forecasts U.S. data center power demand climbing from 31 gigawatts in 2025 to 41 gigawatts in 2026 and 66 gigawatts in 2027—a trajectory that would require adding the equivalent of multiple large power plants every single year.

The bottleneck is not generation capacity per se; it is interconnection. Utilities are overwhelmed with connection requests, and the queue to get a new data center onto the grid has stretched to five, seven, or even ten years in many major markets. For hyperscalers and AI infrastructure developers who need to bring facilities online in 12 to 24 months to meet customer demand, that timeline is functionally impossible.

This is the gap Bloom Energy and Brookfield are betting they can fill. Bloom’s solid oxide fuel cell systems can be installed on-site within months, entirely bypassing grid interconnection queues. They generate electricity directly from natural gas or hydrogen through an electrochemical reaction rather than combustion, making them quieter, cleaner, and more efficient than traditional backup generators or peaker plants.

Inside the $25 Billion Framework

The partnership—originally announced in October 2025 at a $5 billion commitment—has grown dramatically as both parties observed the acceleration of AI infrastructure spending and validated early project economics. The expanded framework will be deployed through Brookfield’s dedicated AI Infrastructure Fund, which launched in November 2025 with a target to deploy $100 billion into AI-related assets globally.

Brookfield, which manages over $1 trillion in assets across infrastructure, real estate, renewable energy, and private equity, is positioning itself as an end-to-end AI infrastructure financier—able to provide not just capital, but operational scale across power, compute, and data center construction.

“This partnership expansion reflects our conviction behind our broader AI infrastructure strategy,” said Sikander Rashid, Head of AI Infrastructure at Brookfield, “and strengthens our position to deliver end-to-end solutions, from electrons to tokens.” The phrase is instructive: Brookfield is not thinking of itself merely as a power provider but as an integrated AI factory builder.

Bloom Energy’s Chief Commercial Officer Aman Joshi was equally direct about the market pull: “Bloom is uniquely positioned to address the urgent need for clean, reliable power to support AI growth,” noting that recent momentum reflects “large-scale deals” being signed across the hyperscaler landscape.

Why Fuel Cells Over Other Alternatives?

The AI industry has explored nearly every available power option in response to grid constraints. Nuclear advocates have pointed to small modular reactors as a long-term solution, but SMRs remain years from commercial deployment at scale. Natural gas peaker plants can be built relatively quickly but face permitting challenges and environmental opposition. Solar and battery storage can work for certain workloads but cannot reliably deliver the continuous, high-density power that GPU-dense AI training clusters demand.

Fuel cells occupy an interesting niche. Bloom’s solid oxide technology runs on natural gas (or, eventually, hydrogen) and achieves electrical efficiencies of up to 65%—roughly twice the efficiency of a conventional gas turbine. Because they operate without combustion, they produce no NOx, SOx, or particulate emissions that would trigger air quality permits in most jurisdictions. A fuel cell installation that would face years of environmental review if it were a gas-fired power plant can often be sited and approved in a fraction of the time.

The trade-off is cost. Fuel cells remain more expensive per kilowatt-hour than grid power in most markets. But that calculus changes when the alternative is a five-year wait to connect to the grid, and when the AI workloads running on that power generate thousands of dollars per GPU-hour in revenue.

The Emerging Off-Grid Data Center Model

The Brookfield-Bloom deal is the largest single manifestation of a broader structural shift: AI data centers are increasingly being designed and financed as self-contained energy systems rather than buildings that plug into the grid. This mirrors an evolution that occurred in telecommunications in the 2000s, when carriers built out independent fiber networks rather than depending on incumbent telco infrastructure.

Several major hyperscalers have already signed long-term power purchase agreements with fuel cell providers or have built private gas pipelines to data center campuses. Microsoft, in particular, has been associated with Bloom Energy deployments at several facilities, and the Brookfield framework is expected to accelerate installations at Microsoft-adjacent infrastructure—though specific customer relationships were not named in the announcement.

The broader implication is that AI infrastructure is becoming a vertically integrated stack: compute, cooling, power generation, and capital all bundled into a single developer relationship. Brookfield’s positioning at the top of that stack—providing both the financing and the operating model—suggests that asset management firms are becoming as important to the AI buildout as the chip companies and cloud providers that generate most of the headlines.

Looking Ahead

The $25 billion framework is a commitment, not a deployment figure—the capital will be drawn down as individual projects are identified, permitted, and constructed. The speed at which Brookfield can deploy it will depend on how quickly the AI infrastructure market continues to absorb capacity, and on whether Bloom Energy can scale its manufacturing to match.

For investors and policymakers watching the AI infrastructure build, the deal is a data point about where the smart money sees the critical path. The constraint is not compute; NVIDIA and its competitors are adding GPU capacity rapidly. The constraint is electrons. And the companies solving the electron problem at scale are, increasingly, where the action is.