Bloom Energy positioned as primary distributed power supplier for hyperscale AI data centers; outpaces nuclear SMR timeline.
The artificial intelligence boom has shifted from a race for chips to a race for electricity. Every new AI data center requires enormous amounts of reliable power, yet the U.S. electric grid is struggling to keep pace. This mismatch is becoming one of the defining investment themes of the decade.
While semiconductor companies remain at the center of AI spending, the companies capable of supplying power quickly may become just as important. Hewlett Packard Enterprise's latest outlook illustrates just how large the challenge has become and why Bloom Energy appears uniquely positioned to benefit.
At HPE's 2026 Discover IR Summit, CEO Antonio Neri cited a critical statistic: the U.S. is on track to face a 19-gigawatt power gap by 2028—enough electricity to power roughly 16 million homes. Neri added that data centers could account for nearly half of U.S. electricity demand through 2030. These numbers explain why "time to power" has become almost as valuable as computing power itself.
The power shortage creates opportunities across several industries, but the most compelling lies with behind-the-meter solutions deployable in months. Unlike utilities building billion-dollar power plants, Bloom Energy installs modular solid oxide fuel cell (SOFC) systems directly at customer sites. Operating primarily on natural gas, the systems generate electricity where it's consumed, reducing dependence on an overloaded grid.
Speed is critical. According to Bloom Energy's 2026 Data Center Power Report, 27% to 38% of data centers are expected to rely on onsite power either fully or partially by 2030, a sharp increase from current adoption. Many Bloom systems can be deployed in roughly 90 days, compared with years for new grid-scale generation. Morgan Stanley recently estimated onsite solutions like Bloom's could contribute 5 GW to 8 GW toward closing America's growing power deficit.
Customers are already signing large agreements. Bloom expanded its master services agreement with Oracle to support up to 2.8 GW of capacity, with 1.2 GW already under contract. The company signed a 20-year, $2.65 billion agreement with American Electric Power for up to 1 GW, alongside a $5 billion strategic partnership with Brookfield Asset Management targeting AI infrastructure. Reports indicate Bloom secured approximately $7.65 billion in data center-related contracts over a relatively short period.
Bloom's operating performance suggests those contracts are beginning to translate into revenue. First-quarter earnings show revenue climbed 130% year over year to $751 million, while product revenue surged 208%. Management increased full-year 2026 guidance to $3.4 billion to $3.8 billion, implying approximately 80% growth at the midpoint following record $2.02 billion in 2025 revenue.
Risks remain. Bloom must successfully expand manufacturing from roughly 1 GW of annual production capacity to 2 GW by the end of 2026 while managing fuel availability, permitting requirements, and a valuation that assumes continued execution. However, these risks appear more manageable than waiting years for traditional power plants to come online.
AI's next bottleneck isn't chips—it's electricity. HPE's projection of a 19 GW power shortfall by 2028 demonstrates just how urgent the challenge has become. While GE Vernova should benefit over the long run as utilities build new generating capacity, those projects require years to complete. Bloom Energy operates on a different timeline. Its modular fuel cells can be deployed in months, its customer list already includes Oracle, American Electric Power, and Brookfield Asset Management, and its revenue is already accelerating. When the market needs power today instead of five years from now, Bloom Energy represents one of the clearest investment opportunities in the AI infrastructure buildout.