# BE How the company works

> How Bloom Energy (BE) works, in its own words from its annual report for 2025: how it makes money, what it sells, unit economics, competitive position, cost inputs, and the anatomy of the extremes: what has to break for the result to land far from the forecast, either way.

Canonical: https://pricedin.markets/be/company/

## How the company works

What it sells, how it makes money, and the anatomy of the extremes: what has to break for the result to land far from the forecast, either way. In Bloom Energy's own words, from its annual report for 2025.

Bloom Energy makes solid-oxide fuel cells that turn natural gas into electricity on the customer's own site, without combustion, and sells them to data centers, utilities and factories that cannot wait for the grid.

#### How it makes money

Mostly the sale of Energy Servers, then years of service. Product revenue was $1,531.3M in 2025, up 41.1%, installation $204.1M, service $228.3M and electricity sold under power purchase agreements $60.4M: total revenue $2,024.0M, up 37.3%, at a gross margin of 29%. Income from operations was $72.8M and the net loss attributable to common stockholders $88.4M. Three customers were 43%, 13% and 12% of revenue, the first a related party, SK ecoplant.

#### What it sells

| 2025 | Revenue | Change |
| --- | --- | --- |
| Product Bloom Energy Server fuel cell systems; the Bloom Electrolyzer for hydrogen | $1,531.3M | +41.1% |
| Installation | $204.1M | +66.8% |
| Service long-term operations and maintenance contracts on the installed base | $228.3M | +6.9% |
| Electricity power sold from systems Bloom owns under power purchase agreements | $60.4M | +14.2% |

Revenue by line as Bloom reports it. Gross margin: product 35%, installation −1%, service 10%, electricity 46%.

#### Unit economics

A factory business scaling up: the Fremont, California cell-printing plant and Delaware assembly ran at about 1 GW a year, and the company plans to double that to 2 GW by the end of 2026. Research is $186.0M a year; 2,214 employees. Sales are large, lumpy and slow to close, and the customer often needs a financier: a $5.0B framework with Brookfield over five years, and SK ecoplant as channel and financing partner in Korea, where nearly 682 MW is deployed. Service contracts on about 1,100 sites in 9 countries are the recurring line, profitable since 2025 at a 10% gross margin.

#### Competitive position

The company competes "against alternative sources of electricity generation which provide firm, always on power": the utility grid first, then gas reciprocating engines and turbines from the established equipment makers, whose lead times "span multiple years". Its case: a modular system that can be delivered and running while an interconnection queue, 2,300 gigawatts long at the end of 2024, waits; higher efficiency and lower emissions than engines; and a response to AI load steps "at least twice as fast as turbines and engines". Its partners name the market: American Electric Power, up to 1 GW for AI data centers; Oracle, on-site power for AI data centers; Brookfield.

#### Biggest cost inputs

The cells and stacks it prints itself and the balance of plant it buys, including components from single or limited sources; factory expansion to 2 GW; installation labor and site work; warranty and service on the fleet; and interest on its debt. The customer's own cost is natural gas, and the federal investment tax credit for fuel cells on gas expired at the end of 2024, with qualifying transactions carrying some projects to 2028.

#### Anatomy of a large **upside**

- **AI data centers make on-site power the default.** Grid queues and multi-year turbine lead times push hyperscalers and their developers to buy power they can install now; the AEP, Oracle and Brookfield agreements turn into gigawatt orders and the 2 GW factory fills.
- **Scale and cost down turn into profit.** Product revenue grew 41.1% and operations turned to income of $72.8M; a second gigawatt of capacity on the same plant, and service now profitable, bring the margin and net income the business has not yet shown.
- **Hydrogen and new markets open.** The Bloom Electrolyzer and the same platform on hydrogen, biogas or carbon-captured gas, plus Europe and Asia growing "from a smaller base", give the plant markets beyond the US data center.

#### Anatomy of a large **downside**

- **The customer base is three names and a credit.** 43%, 13% and 12% of revenue came from three customers; sales involve "a lengthy sales and installation cycle" and depend on financing partners, so a lost or delayed order moves the year.
- **Gas and incentives turn against it.** "Anti-natural gas sentiment", renewable procurement goals, a rising gas price or the loss of remaining tax incentives after the fuel cell investment tax credit expired cut the economics against the grid and against engines.
- **The incumbents catch up or the rival technologies win.** Turbine and engine makers add capacity and shorten lead times, or "small scale nuclear and geothermal" and batteries take the data center load, and a 2 GW factory built for the boom runs half full.

Written from Bloom Energy's annual report for 2025, in its words and numbers. The extremes are the long view, what has to break for the result to land far from the forecast; the quarter's own cases, each model's, are on the quarter forecast.