Nebius is partnering with Bloom Energy to deploy solid oxide fuel cell systems for AI infrastructure sites in the United States, as AI cloud providers race to secure power capacity amid growing grid constraints. The first deployment is expected to deliver 328 MW of installed capacity this year, replacing previously planned gas turbine infrastructure with Bloom’s combustion-free fuel cell technology.
The agreement positions Nebius among a growing group of AI infrastructure operators turning to onsite power generation to accelerate data center deployment timelines. Bloom’s modular fuel cell systems will operate behind the meter, reducing dependence on utility transmission upgrades and helping Nebius shorten time-to-power for AI training and inference clusters. The companies said the systems offer high efficiency, low emissions, and reduced water consumption compared to conventional combustion-based generation.
Nebius said the partnership supports its broader expansion strategy across the U.S. and EMEA regions, where the company is building AI-native cloud infrastructure for model development, training, and deployment. The agreement also includes the potential for future international expansion as Nebius scales additional AI compute sites globally.
• First U.S. deployment targets 328 MW of installed onsite power capacity
• Fuel cells will replace previously planned combustion-based gas turbine systems
• Behind-the-meter deployment reduces dependence on transmission expansion
• Bloom fuel cells generate electricity without combustion and with minimal water use
• Partnership supports Nebius AI infrastructure expansion across the U.S. and EMEA
• Long-term agreement includes potential global expansion opportunities
“Power remains a key constraint for AI infrastructure build-outs,” said Andrey Korolenko, Chief Product and Infrastructure Officer at Nebius. “We chose Bloom because their fuel cells solve that directly: Clean power with virtually no pollutants is deployed onsite, on the timelines our customers need, with the availability AI workloads require.”
🌐 Analysis: The AI infrastructure market increasingly treats power availability as a gating factor for GPU cluster deployment, particularly in the United States where utility interconnection queues and transmission upgrades can delay projects by several years. Behind-the-meter generation strategies are becoming more common among hyperscalers, colocation operators, and AI-native cloud providers seeking faster deployment timelines for high-density AI workloads.
🌐 Bloom Energy continues to expand its presence in AI and data center infrastructure markets as demand rises for alternative power architectures capable of supporting multi-hundred-megawatt campuses. The deal also reflects broader industry interest in modular onsite generation technologies, including fuel cells, small modular reactors, and natural gas microgrids, as operators seek scalable approaches to powering next-generation AI factories.
