BE
Bloom Energy
Summary
What they do:
Manufactures solid oxide fuel cells (SOFCs) that convert natural gas directly into electricity on-site, bypassing the grid entirely — giving data center operators power in 18-24 months instead of the 5-10 year grid interconnection queue.
Why they matter:
With grid interconnection queues measured in years and hyperscaler demand measured in gigawatts, Bloom is one of the few companies that can deliver behind-the-meter power at scale, making it a critical enabling technology for the AI buildout timeline.
Recent performance:
FY2025 revenue $2.02B (+37% YoY), Q4 revenue $599.3M (+56.7% YoY). Stock at ~$167, up from $16 52-week low — an 887% one-year run. Backlog $20B. Market cap ~$50B.
Our Verdict
The data center power bottleneck's most compelling solution — Oracle's 2.8GW deal validates the technology at scale, but at $50B market cap on $2B revenue and thin margins, the stock prices in flawless execution of a manufacturing ramp that has never been done before in fuel cells.
Structural trends
Structural
66
/ 100
Moat
6/10
Fuel cells
AI Exp.AI Exposure
High~70% AI
Play Type
EmergingAI Growth
~80-100%
Rel. Value
45
FAIRPriceLIVE
$219.03
+23.98%
Live via Yahoo Finance · refreshes every 5 min
Market Cap
$62.2B
P/E Ratio
N/A
P/S Ratio
30.8x
52W High
$219.27
52W Low
$16.01
52W Chg
1268.1%
Beta
3.19
Every hyperscaler building AI data centers today faces the same constraint: power. Not computing power — electrical power. Grid interconnection queues in the United States now stretch 5-10 years in most regions, meaning a data center approved today might not have utility power until 2031 or later. Bloom Energy offers an alternative path: install solid oxide fuel cells on-site that convert natural gas directly into electricity at 60%+ efficiency, delivering megawatts of power in 18-24 months.
The technology is elegantly simple in concept, brutally difficult in execution. Bloom's fuel cells use a ceramic electrolyte (yttria-stabilized zirconia) to electrochemically convert fuel to electricity without combustion. This produces roughly half the carbon emissions of a gas turbine per kWh, zero NOx or SOx, and requires no water for cooling — a meaningful advantage in water-scarce regions. Each Bloom Energy Server generates approximately 300kW; data center deployments stack hundreds of these units to reach 50-100MW+ scale.
The company was founded in 2001 by K.R. Sridhar, a former NASA researcher who developed fuel cell technology for the Mars program. After 17 years of R&D and gradual commercial deployment, Bloom went public in 2018 at $15/share. For most of its public life, the stock traded between $10-$30, serving a modest base of commercial and industrial customers. Then the AI data center boom arrived.
FY2025 was a transformation year. Revenue reached $2.02B (+37% YoY), with Q4 alone generating $599.3M (+56.7%). The company secured its landmark deal with Oracle — a 2.8GW framework agreement that alone could represent $8-10B in revenue over multiple years. Total backlog stands at $20B, providing roughly 10x trailing revenue in contracted work. Management has guided FY2026 revenue of $3.1-3.3B (+50%+ YoY) with further acceleration expected in FY2027.
Bloom operates primarily from its manufacturing facility in Fremont, California, with a newer facility in Newark, Delaware (the former Chrysler plant). Current manufacturing capacity is approximately 2GW annually, with plans to expand as demand requires. The company has approximately 3,000 employees.
The business model is shifting. Historically, Bloom sold fuel cell systems outright (hardware sales) with long-term maintenance contracts. Increasingly, the company is moving toward power-as-a-service (PaaS) arrangements where Bloom owns the fuel cells, installs them at the customer site, and charges per kWh over 10-20 year contracts. This creates recurring revenue but requires significant upfront capital. In FY2025, product revenue was $1.29B with service revenue of $728M.
Supply Chain Dependencies
Upstream Suppliers
Downstream Customers
The Catch
Bloom Energy is a $50B company generating $2B in revenue with thin-to-negative GAAP earnings. The 887% one-year stock run — from $16 to $167 — has priced in a manufacturing ramp that has never been accomplished in fuel cell history. The Oracle 2.8GW deal is transformative but it's a framework agreement, not a firm purchase order, and actual deployment pace depends on Oracle's construction timeline. More fundamentally, fuel cells face competition from gas turbines (proven, cheaper per MW) and nuclear SMRs (zero-carbon, longer timeline), meaning Bloom's window of competitive advantage may be 5-7 years, not 20. At 25x sales on a hardware manufacturing business with 25% gross margins, the valuation leaves zero room for execution stumbles, competitive surprises, or macro-driven capex slowdowns.
If They Win
If Bloom executes the manufacturing ramp, converts the Oracle backlog on schedule, signs additional multi-GW hyperscaler deals, and expands gross margins toward 35%+, the company becomes the on-site power utility for the AI economy — not a fuel cell manufacturer but a distributed power platform that every data center operator contracts with as a prerequisite for construction. Revenue compounds to $6-8B by 2028, margins expand as scale economics kick in, the PaaS model creates a recurring revenue base worth a premium multiple, and hydrogen/electrolyzer optionality adds a second growth vector. In that scenario, the current $50B market cap looks reasonable against a $100B+ revenue trajectory, and Bloom joins the ranks of infrastructure companies (like Eaton or Quanta) that became essential to the AI buildout. The difference: Bloom would be doing it at software-like growth rates.
Others in Secure the Power
Not financial advice. All scores generated via AI algorithms using public data.