TodayToday's edition
Bloom Energy's Q3 revenue rose 138% to $1.24 billion, Reuters reports, as US grid waits of three to six years push data centers toward on-site fuel cells; Rystad puts the order book near 9 GW.
3 min readAI-assisted
A Bloom Energy Server cabinet, the company's solid-oxide fuel-cell system. The photo is a product detail from Bloom's media resources, not an image of Oracle's Utah site.
Image: Bloom Energy
Source · Press image from the Bloom Energy newsroom, used for news reporting.
Bloom Energy's third-quarter revenue rose 138% to $1.24 billion, and the company raised its 2026 revenue forecast to $4.4 billion to $4.6 billion, as AI data centers turn to on-site fuel cells, according to a Reuters explainer published on October 23. Net profit was $250.3 million against a loss of $23.1 million a year earlier.
The figure sits inside a broader shift. Reuters reports that US hyperscalers are turning to fuel cells because grid constraints make it hard to secure quick, reliable power. Energy research firm Rystad Energy estimates that linking large loads to the US grid can take three to six years, and that about 40% of projected US data-center capacity in 2030 could pursue dedicated on-site power instead of relying entirely on the grid.
Oracle is the largest single buyer in the Reuters account. It intends to buy up to 2.8 GW of Bloom fuel-cell systems, of which 1.2 GW is already contracted; that framework predates this quarter, and Reuters gives no new terms for it.
Bloom CEO KR Sridhar said on the post-earnings call that the company delivered power to Oracle's Utah site in less than two months. He said Bloom is doing the same with Nebius in New Jersey, where planning board approvals for the fuel cells came within three months of signing the contract. Those are company claims, not independent measurements.
Fuel cells make electricity through a chemical reaction between a fuel such as natural gas or hydrogen and oxygen, without burning it. Reuters notes that this gives lower local emissions than conventional gas-fired generation. Bloom uses solid-oxide technology ready for 100% hydrogen, while FuelCell Energy uses molten carbonate technology capped at 40% hydrogen blends. Both can run on other fuels.
FuelCell Energy CEO Jason Few told Reuters that time to power has become as important as cost of power. The company expects data centers and related business to make up 75% to 85% of its business within two to three years, up from 25% to 30% now.
For power planners, the useful comparison is scale. Rystad puts the industry order book at roughly 9 GW, including framework agreements involving Oracle, AEP, Equinix and Brookfield. Rystad expects annual fuel-cell manufacturing capacity to rise to 4 GW by 2030 from about 1.8 GW today, and projects investment in data-center fuel cells of about $30 billion by 2030, up from $2.8 billion in 2025. On those figures, Oracle's 2.8 GW alone is large against current yearly output, so delivery would stretch over several years.
Rystad expects annual fuel-cell manufacturing capacity to rise to 4 GW by 2030 from about 1.8 GW today, according to Reuters.
Illustration (AI-generated)
Behind-the-meter fuel cells also move the bottleneck rather than remove it. Bloom Chief Commercial Officer Aman Joshi told Reuters that for projects of 100 MW and above, timelines depend mainly on site readiness, permitting and supporting infrastructure, not the fuel-cell technology. Reuters adds that projects still need gas connections, permits, transformers and switchgear. Voltgrid News reported separately on INNIO's Jenbacher gas engine order from a US data-center developer.
Reuters says framework agreements do not necessarily represent firm orders for the full stated capacity, so the 9 GW order book is an estimate rather than booked supply. Of Oracle's 2.8 GW, 1.6 GW is not yet contracted. The report gives no price, contract term or site list for the Oracle volume. Reuters says growth depends on whether suppliers can deliver at competitive cost and whether developers can build projects and secure fuel.
