Fuel cell technology is emerging as the unexpected solution to AI's insatiable appetite for electricity. As hyperscalers like Microsoft, Alphabet, Meta Platforms, and Oracle pour billions into next-generation artificial intelligence (AI) data centers, they're running into a critical problem: traditional electrical grids simply cannot keep up. Rather than waiting years for grid expansions or new nuclear plants, these tech giants are turning to on-site fuel cell systems that can be deployed and operational in as little as 90 days.

Bloom Energy, a company specializing in solid oxide fuel cells, has become the primary beneficiary of this shift. The company's stock has surged more than 250% so far in 2026, driven by concrete commercial contracts rather than speculation. In the first quarter of 2026, Bloom reported record revenue of $751.1 million, a 130.4% year-over-year increase, and swung to profitability with net income of $70.6 million compared to a loss of $19.1 million in the same quarter a year earlier.

How Do Fuel Cells Solve the Data Center Power Problem?

Bloom's solid oxide fuel cells work by converting renewable natural gas, biogas, or hydrogen into electricity without combustion and with minimal carbon dioxide emissions. The key advantage for data center operators is speed and independence from grid infrastructure. By generating electricity on-site, hyperscalers can bypass the traditional power grid bottleneck entirely, ensuring their expensive AI chips don't sit idle waiting for power to become available.

• Rapid Deployment: Fuel cells can be installed and operational in as little as 90 days, compared to years required for grid expansion or nuclear plant construction.
• Renewable Energy Source: Systems use renewable natural gas, biogas, or hydrogen rather than fossil fuels, aligning with corporate sustainability goals.
• On-Site Generation: Eliminates dependence on traditional electrical grids, which are severely bottlenecked by AI infrastructure demand.
• Minimal Emissions: Fuel cells convert energy without combustion, producing minimal carbon dioxide emissions compared to traditional power generation.

What Are the Major Contracts Driving Bloom's Growth?

Bloom's backlog tells the story of AI's power crisis more clearly than any analyst report. As of the end of 2025, the company reported a backlog of $20 billion, including a product backlog of $6 billion. These are not speculative orders; they represent binding commitments from some of the world's largest technology companies.

In April 2026, Bloom expanded its agreement with Oracle to support up to 2.8 gigawatts (GW) of fuel cell capacity. This includes Project Jupiter, a massive multigigawatt AI data center campus in New Mexico that will run entirely on Bloom fuel cells rather than traditional gas turbines or diesel generators. In May 2026, Bloom secured a 328-megawatt (MW) deployment deal with AI infrastructure company Nebius, providing multiyear visibility for revenue generation.

These contracts represent a fundamental shift in how hyperscalers approach infrastructure. Rather than waiting for grid upgrades, they are building self-sufficient power ecosystems using fuel cell technology. The scale of these deployments underscores just how urgent the power bottleneck has become for companies racing to build AI capabilities.

Is Bloom Breaking the Fuel Cell Industry's Profitability Problem?

Historically, fuel cell companies have struggled to turn a profit despite rising revenue. Bloom appears to be breaking that pattern. The company's gross margin expanded beyond 30% in the first quarter of 2026, driven by manufacturing-scale benefits. Earnings per share came in at $0.23, compared to a loss of $0.10 in the same quarter a year earlier, while adjusted earnings per share reached $0.44.

These results were strong enough to prompt management to raise its full-year revenue guidance to between $3.4 billion and $3.8 billion, representing an 80% increase at the midpoint. Adjusted earnings per share guidance was raised to between $1.85 and $2.25, up 170% at the midpoint. The company's ability to generate positive cash flow while scaling production suggests that fuel cell technology may finally be reaching the inflection point where it becomes economically viable at scale.

Bloom does face challenges. The company will need to double its factory capacity from 1 gigawatt to 2 gigawatts by the end of 2026 to meet demand. It also faces competition from Plug Power and FuelCell Energy. However, Bloom's triple-digit revenue growth and positive cash generation distinguish it from competitors who remain trapped in turnaround mode or declining revenue patterns.

What Does This Mean for the Broader AI Infrastructure Landscape?

The rise of fuel cell technology as a solution to AI's power crisis reflects a fundamental reality: the electrical grid was not designed for the power demands of modern artificial intelligence. A single large language model (LLM) training run can consume as much electricity as a small city. As companies deploy increasingly large AI systems, the gap between available grid capacity and required power grows wider each quarter.

Fuel cells represent one piece of the solution, but they are not the only one. Hyperscalers are simultaneously exploring nuclear power, renewable energy sources, and more efficient cooling systems. However, the speed and flexibility of fuel cell deployment make them particularly attractive for companies that cannot afford to wait years for other infrastructure solutions to come online.

The $20 billion backlog at Bloom Energy suggests that this is not a niche solution but rather a core component of how the world's largest technology companies plan to power their AI ambitions. As AI continues to consume more electricity, fuel cell companies that can scale production efficiently will likely see sustained demand for years to come.