Founded in 2001, Bloom Energy’s initial pitch was that in a digitized world, “there’s no such thing as enough energy,” recalled CEO KR Sridhar. And large users like data centers or manufacturers would need that power to be produced onsite, even if they’re relying primarily on the grid.
At a certain point, the fuel cell company began pitching its modular generators as a clean energy solution; even though they ran and still run primarily on fossil gas, their hydrogen compatibility was part of the appeal. Bloom began making green hydrogen electrolyzers in 2020, angling for a stake in the market for decarbonizing power.
However, in recent years, that market has flagged, right as load growth has surged. And Bloom has found itself with an opportunity.
Data centers have long been a major customer segment for Bloom, and are one of the main drivers of the new rise in electricity. As interconnection timelines grow, large power users like hyperscalers are increasingly prioritizing speed to power over clean energy. (While hydrogen remains in short supply, Bloom is relying on carbon capture to keep emissions down: a compromise that many of the biggest players in the data center boom are also grappling with.)
Bloom has already signed major deals with customers like the utility AEP and the data center company Equinix; last November, just two days after President Trump’s election, Bloom announced plans to create the world’s largest fuel cell installation with its Korean partner and distributor SK Eternix.
But the opportunity isn’t straightforward. It’s a hectic and competitive market, made more so by a moment of policy and economic uncertainty.
Bloom reached “unicorn” status with $1.5 billion in investment before IPOing in 2018. However, the years since have been a bit of a rollercoaster. Last week, Bloom reported $326 million in revenue for the first quarter of 2025, or nearly a 39% year-over-year increase; the better-than-expected earnings was partially due to quirks of project timing. Previous quarters have been less rosy, and the company’s stock plunged in recent months.
And even 24 years in, though the company has had some good quarters, Bloom has only ever posted an operating profit once, in 2024. The company still has never had a net profitable year.
On the same earnings call, the company also announced that CFO Dan Berenbaum would depart after just a year on the job.
Analysts have had a mixed reaction in the days post-earnings. The Japanese financial group Mizuho upgraded Bloom Energy, noting that “we could see more deals with large load customers extending until 2040.” But Truist Bank kept a hold on the shares, telling investors that tariffs and macroeconomic uncertainty are causing headwinds, as is the fact that artificial intelligence models are becoming more energy efficient.
That said, Sridhar’s tone was optimistic on the company’s earnings call, saying the “gigawatt gap” is so large that Bloom continues to project growth: “This is an investment super cycle, and short-term economic issues will not adversely impact the megatrend.”
The power plant in a box, for data centers
Bloom unveiled its “power plant in a box” in 2010, to an audience that included tech and energy executives, and then-Governor Arnold Schwarzenegger. It was an era of enthusiasm for fuel cells for everything from cars to power systems, and Bloom specifically promised that its hydrogen-compatible modular system could be part of a clean energy overhaul for data centers and commercial buildings.
In the years since, hydrogen — and especially green hydrogen produced by electrolysis — has largely failed to take off commercially. And so Bloom has pivoted, to a system that relies instead on both gas and carbon capture. The energy servers use solid oxide fuel cells, which Sridhar noted “is the only technology that gives you a very concentrated tailpipe of carbon dioxide.”
“We do carbon capture automatically,” he added. “We believe that using natural gas to provide these large loads of power, capturing the carbon dioxide and sequestering it, is one of the best ways for decarbonization.”
The systems have the ability to switch between hydrogen, biogas, fossil gas, or a blend. “If hydrogen were available at a large scale and at an economically viable cost, then we would switch immediately,” Sridhar told Latitude Media. “But it is not.”
That said, today’s customers may simply not care. Both energy companies and data center developers are increasingly cutting deals for new gas generation, especially to serve artificial intelligence demand. There’s a sense that while hyperscalers with ambitious decarbonization timelines would like to use clean energy, the priority is to get any power at all, and fast.
Other companies, including the Stanford offshoot Mainspring Energy, are also pushing fuel-agnostic generators for data center customers. Crusoe Energy, which brings both (largely gas) power generation and data center development under one roof, is fast becoming a major player in the AI boom.
And the Trump administration has made it clear that gas is a favored fuel, from reopening federal lands and waters to exploration and extraction, to removing restrictions on certain LNG exports.
For more on the potential of CCS for data centers, listen to Julio Friedmann, chief scientist at Carbon Direct, on Catalyst with Shayle Kann:
In this context, Bloom is emphasizing reliability. The question, Sridhar said, is “how do we adapt to the varying load of a data center and give them that islanded solution where they don’t have to worry about that load going up and down, and they don’t have to worry about having a backup of the entire capacity backed up by the grid?”
The company’s customer segments are split three ways. Roughly 40% are commercial and industrial customers with high reliability requirements, like hospitals or telecommunication; 30% are international customers, especially in Korea; and the final 30% are, and have always been, data centers.
“The demand growth here [in the U.S.] is faster and more hectic than we have ever seen before,” Sridhar said. “And that’s both data centers and non-data centers.”
While the data center demand is fairly straightforward, Bloom has observed that there’s a sense of competition that breeds demand from other large, non-data center power users. “The non-data center guys feel, ‘If data centers can afford to write the big checks and get the complete attention of the utility industries, we are going to be left behind,’” Sridhar said. “So they’re trying to secure their own power.”
This coincides with a moment of enormous market uncertainty, pegged largely to the Trump presidency and his erratic approach to policy-making. Tariffs, for instance, are already complicating domestic solar and storage manufacturing plans.
Sridhar, though, seems unconcerned. The company makes its systems in the U.S., with two domestic manufacturing and assembly plants. “We have been developing, diversifying, and fortifying our supply base for years to mitigate the impacts of any particular country or supplier,” Sridhar said. “While we import material and components from abroad, we do not depend on China.”
