Bloom Energy and Brookfield Asset Management announced a joint plan to invest $5 billion to build power and data infrastructure aimed at large AI workloads. The deal pairs Bloom’s onsite fuel cell systems with Brookfield’s project development and financing muscle. They aim to create integrated “AI factories.” These will be sites that combine servers, cooling, and local power into one optimized system.
The partnership aims to solve two linked problems. First, AI data centers need huge amounts of steady power. Second, grid upgrades often lag behind where compute demand grows fastest. By putting low-emission power on-site, the partners hope to cut delivery delays and lower operating risk for large AI customers.
KR Sridhar, Founder, Chairman and CEO of Bloom Energy, remarked:
“Unlike traditional factories, AI factories demand massive power, rapid deployment and real-time load responsiveness that legacy grids cannot support. The lean AI factory is achieved with power, infrastructure, and compute designed in sync from day one. That principle guides our collaboration with Brookfield to reimagine the data center of the future. Together, we are creating a new blueprint for powering AI at scale.”
The $5B Power Play Behind the AI Boom
Brookfield will provide capital in stages to fund the deployment of Bloom Energy fuel cells at AI data center clusters. Bloom will supply, install, and maintain the fuel cell systems, Bloom’s Solid Oxide Fuel Cell (SOFC), and work with Brookfield on site design.
The two firms will co-develop sites in North America, Europe, and other regions. One pilot location in Europe is already in early development, with more sites planned as the program scales.
Bloom’s fuel cells run on a range of fuels, including natural gas today and hydrogen or biogas in low-carbon scenarios. The systems generate power behind the meter. That means the power is made and used on-site. On-site generation cuts reliance on long transmission lines. It also speeds up project timelines, unlike waiting for big grid upgrades.
Brookfield will target locations with constrained grids or high energy costs. It will combine finance, real estate, and engineering to deliver turnkey AI campuses. Bloom will focus on power technology and operations. The joint model aims to sell access to compute capacity bundled with resilient, lower-emission power.
Each megawatt of Bloom Energy’s fuel-cell power can help avoid about 4,000 tons of CO₂ emissions per year when it replaces diesel generators. Fuel cells using renewable hydrogen can achieve net-zero emissions. This provides a cleaner option for energy-intensive AI infrastructure.
Over time, the company expects to make an even greater impact. Bloom’s long-term goal is to cut millions of tons of CO₂ annually across its growing network of AI campuses and data center projects.
AI Finding Its Own Power Source
AI compute growth is moving fast. Some industry estimates say U.S. AI data center demand could exceed 100 gigawatts by the mid-2030s. Global demand for compute and associated cooling and power could reach several times that level.
Hyperscale data centers still put heavy demands on local grids. This often leads to long waits for interconnections.
Data center power is measured in megawatts per facility. Large AI sites can require tens to hundreds of megawatts. For comparison, a 100-MW cluster needs roughly the same continuous power as a small city.
If many new AI sites come online in the same region, the grid must expand quickly. That expansion often takes years. On-site fuel-cell power can provide an interim or long-term solution in such cases.
Analysts value AI infrastructure as a major growth market. Some forecasts estimate that the AI infrastructure opportunity will reach trillions of dollars in the next decade. This includes costs for servers, cooling, power, and facilities. The $5 billion partnership is one of the earliest large, purpose-built plays aimed directly at that market.
Wall Street Takes Notice as AI Energy Heats Up
Markets reacted strongly when the deal was announced. Bloom Energy shares jumped in early trading. This rise shows that investors believe the company can land long-term orders from AI operators. Analysts raised revenue forecasts for Bloom based on expected project pipelines tied to AI data centers.
Brookfield’s move fits a wider trend of asset managers investing in digital infrastructure. These investors see steady, long-term cash flows from data center leases and embedded power contracts. The partnership blends that capital with a technology supplier that can deliver power where it is needed.
Economics relies on several factors: fuel prices, local power rates, incentives for low-carbon energy, and the costs of installing and running fuel cells at scale. If hydrogen or other low-carbon fuels fall in price, the climate benefits of onsite fuel cell power grow. If local rules penalize behind-the-meter generation, projects may need different commercial structures.
Can Fuel Cells Handle the AI Load?
The plan has real technical and market risks. Fuel cells must prove long-term reliability at the scale AI factories require. These systems also need supply chains for parts and fuels.
Project teams must integrate power with cooling, backup systems, and server infrastructure. That requires careful engineering and long maintenance cycles.
Regulatory and permitting rules vary by country and by city. Some utilities and regulators are cautious about large onsite generators. In some markets, onsite generation faces higher charges or must follow strict interconnection rules. The partners will need to adapt to local rules and often negotiate with utilities.
Another risk is demand timing. AI compute growth could slow or centralize differently than current forecasts assume. If demand grows more slowly, some planned projects could face lower returns. Conversely, very rapid demand could strain component supply chains and raise costs.
Clean Power Meets Compute: The Policy Advantage
Fuel cells offer lower local emissions compared with diesel generators and can reduce grid congestion. When paired with low-carbon fuels such as green hydrogen or biogas, they can cut lifecycle emissions further. The partners say they will pursue lower-carbon fuels as markets and supplies mature.
Policy incentives and carbon pricing will matter. Regions that reward low-carbon onsite power will make the business case stronger. Where grids decarbonize rapidly, the marginal benefit of onsite fuel cells shifts. The partners will likely target places where grid constraints and carbon rules create the greatest value.
The deal also signals a shift in how infrastructure is designed. Rather than treating power and compute as separate systems, the AI factory model integrates them. That can boost efficiency, but also concentrates physical and regulatory risk in single sites.
The Blueprint for Tomorrow’s AI Factories
If Bloom and Brookfield execute well, they could set a new standard for AI infrastructure. The model could scale to dozens of sites and to hundreds of megawatts of deployed fuel-cell power over time. That would create a steady pipeline of orders for Bloom and steady cash flows for Brookfield-managed projects.
The partnership shows how private capital and specialized technology can combine to solve urgent infrastructure gaps. It also shows the complexity of the energy transition. New power sources, fuels, and commercial models need to fit with local rules and physical grids.
AI operators can benefit from bundled offers. These include compute power along with resilient, lower-emission energy. This can shorten build times and lower long-term risks. For investors, the play is a bet on both AI demand and the economics of onsite, low-carbon power.
The $5 billion Bloom-Brookfield partnership aims to knit together power and compute in the AI era. It responds to a clear need: massive, concentrated power for AI that sometimes outpaces grid upgrades. The move could accelerate new site builds and show a practical path to combine finance, power technology, and data center design.
