Bloom EnergyThin moat

Bloom Energy sells electricity-making machines. Its Energy Server is a stack of solid-oxide fuel cells — ceramic plates that convert natural gas, biogas or hydrogen into power through an electrochemical reaction rather than combustion — installed on a customer's own site, behind the meter, running around the clock. The pitch has always been the same: power that arrives when the grid cannot deliver it, at a premium price the customer pays for speed and reliability rather than for cheapness.

The revenue splits four ways, and the split matters. Of 2025's $2.02 billion¹, about $1.53 billion was Product — selling the boxes — with $0.23 billion of Service (the contracts that maintain them and replace stacks over time), $0.20 billion of Installation and $0.06 billion of Electricity from systems Bloom owns and sells power off. Product is the growth; Service is the annuity that quietly compounds underneath it as the installed base ages.

For most of the company's public life this business did not work. Bloom lost money in every year from its 2018 listing through 2025 — the losses were the business model's honest verdict, not an accounting quirk. Then the AI data-center build-out arrived and turned a niche product into a scarce one, because a data center that cannot get a grid connection for four years will happily buy its own power plant. Trailing revenue is now about $3.11 billion against $2.02 billion for full-year 2025, and the company reached its first profitable trailing year, roughly $0.24 billion².

The market has priced that turn with extraordinary confidence: about $80.4 billion of market value on 294.5 million shares — roughly 26 times sales and, on a first-ever profitable year, something like 328 times earnings³. That is a valuation for a company whose advantages are durable and whose demand is permanent.

Whether either is true is exactly what the following pages test. The Moat examines the fuel-cell technology, the installed base and its service annuity, the AI demand surge and the manufacturing position — and rates the whole thing thin, because a differentiated product that has never earned its cost of capital is not yet a moat. The Future Bets follow the gigawatt contracts with Oracle and AEP, the factory being doubled to build them, and the hydrogen business waiting in the same stack. Its four revenue lines are taken in turn in The Revenue Lines.