Bloom Energy -- How the Business Works

Bloom Energy is the world leader in solid oxide fuel cell (SOFC) power generation, holding an estimated 40-55% share of the commercial stationary SOFC market. The company manufactures Energy Servers that electrochemically convert natural gas (or hydrogen) into electricity at ~60%+ electrical efficiency — no combustion, no turbines. Revenue comes from four streams: Product ($653M, 87% of Q1'26 revenue), which is the sale of Energy Server hardware; Service & Maintenance ($62M, 8%), recurring contracts on the installed base; Installation ($26M, 3%); and Electricity/PPA ($10M, 1%), a declining legacy model. FY2025 revenue totaled $2.0B (+37% YoY) with 30.3% non-GAAP gross margins, $73M GAAP operating income, and a $20B total backlog driven by AI data center power demand outpacing grid capacity.

Q1'26 Revenue

$751M

+130% YoY

Total Backlog

$20B

Multi-year visibility

Non-GAAP Gross Margin

31.5%

+860bps over 4 years

Market Cap

~$68B

NYSE: BE | ~282M shares

Four revenue streams -- Q1'26 breakdown ($751M total)

Q1 2026 Revenue by Segment -- Ranked by Size

Product Revenue

Energy Server hardware sales

$653M (87.0%)

Service & Maintenance

Recurring installed-base contracts

$62M (8.2%)

Installation

On-site deployment services

$26M (3.5%)

Electricity (PPA)

Legacy power purchase agreements

$10M (1.3%)

Segment data from Bloom Energy Q1 2026 10-Q. Source: Daloopa.

The product -- how a Bloom Energy Server works

SOLID OXIDE FUEL CELL

~60%+ Efficiency

Electrochemical, no combustion

Each Energy Server is a modular 300 kW unit containing stacks of solid oxide fuel cells. Natural gas flows across a ceramic cell at ~800°C, triggering an electrochemical reaction that converts fuel directly to electricity — no burning, no turbine, no moving parts. This yields ~60%+ electrical efficiency vs ~35-40% for gas turbines and ~30% for diesel generators. The modular design means capacity scales linearly: a 10 MW data center installation uses ~33 servers deployed in weeks, not years.

SPEED TO POWER

Weeks, Not Years

vs 5-10 yr grid interconnection

The killer value proposition for data centers: grid interconnection queues now stretch 5-10 years. AI hyperscalers need power now. Bloom Energy Servers deploy behind the meter in weeks to months, connecting to existing natural gas infrastructure without grid approval. This is why Oracle committed to 2.8 GW, Brookfield to $25B, and AEP to $2.65B — Bloom solves the time-to-power bottleneck that no gas turbine or grid project can match.

FUEL FLEXIBILITY

Multi-Fuel Ready

Natural gas, biogas, hydrogen

Bloom's SOFC platform runs on natural gas today but is designed for hydrogen compatibility — the same cell chemistry works with H2, producing only water and electricity. Bloom also offers an electrolyzer product that runs the fuel cell in reverse, generating hydrogen from electricity. This dual capability (fuel cell + electrolyzer) positions Bloom for both the current natural gas bridge and the longer-term hydrogen economy transition.

Product specifications from Bloom Energy investor presentations and 10-K filings.

Business model flywheel -- the AI power demand cycle

Step 1 -- AI Creates Power Demand

Hyperscalers Need Gigawatts of New Power, Fast

AI training and inference require massive power. A single large GPU cluster consumes 50-100 MW. Global data center power demand is projected to double by 2030. But grid interconnection queues stretch 5-10 years — the grid cannot keep pace. This structural deficit forces hyperscalers to seek behind-the-meter distributed power solutions that can deploy immediately. Bloom's Energy Servers are purpose-built for this gap.

▼

Step 2 -- Bloom Wins Mega-Deals

$20B Backlog: Oracle 2.8 GW, Brookfield $25B, AEP $2.65B

Bloom's modular deployment model and fuel cell efficiency win contracts that gas turbines and diesel generators cannot fulfill on the required timeline. Product revenue ($653M in Q1'26, +209% YoY) is the primary revenue driver. Each deployment generates a multi-year service contract, creating a growing recurring revenue stream ($62M/quarter) that scales with the installed base. Management raised 2026 revenue guidance to $3.4-3.8B.

▼

Step 3 -- Scale Drives Margin Expansion

Non-GAAP Gross Margins: 21.7% (2021) to 31.5% (Q1'26)

Higher volumes spread fixed manufacturing costs across more units. Product mix is shifting toward higher-margin fuel cell systems as the PPA/electricity segment declines. GAAP operating income turned positive in 2024 ($23M) and scaled to $72M in Q1'26 alone — nearly the full-year 2025 total in one quarter. Management guides 34% non-GAAP gross margins for FY2026 and $600-750M non-GAAP operating income.

▼

Step 4 -- Invest in Capacity

$2.6B Debt Raised to Fund Manufacturing Scale-Up

Bloom raised $1.6B in zero-coupon convertible notes in Q4'25 to fund capacity expansion needed to fulfill the $20B backlog. FCF is currently negative — the company is in investment mode. The bet is that manufacturing scale will drive unit economics to FCF-positive as the backlog converts over 2026-2028. Shares outstanding have diluted 32% over 2.5 years as converts exercise, a real cost to equity holders.

▼

Step 5 -- Installed Base Creates Recurring Revenue

Every Server Deployed Generates 10-15 Years of Service Revenue

Each Energy Server requires ongoing maintenance, cell stack replacements, and monitoring. Service revenue ($228M in FY2025) grows predictably as the installed base expands. This is the flywheel: product sales drive installed base growth, installed base drives recurring service revenue, recurring revenue provides visibility and cash flow stability as the company scales toward FCF-positive. The $20B backlog, if converted, implies a massive expansion of this recurring revenue base over the next 3-5 years.

Flywheel derived from Bloom Energy FY2025 earnings calls, investor presentations, and SEC filings.

End-market exposure -- data centers dominate

DATA CENTERS

~60-70%

Of product revenue (estimated)

Bloom does not disclose end-market revenue splits, but backlog composition tells the story. Oracle (2.8 GW), Brookfield ($25B data center JV), and AEP ($2.65B) are all data-center-linked deployments. AI power demand is the primary growth driver — hyperscalers need behind-the-meter power that deploys faster than grid interconnection allows.

COMMERCIAL & INDUSTRIAL

~20-30%

Hospitals, campuses, manufacturing

The legacy business. Bloom's original market was commercial and industrial facilities needing reliable, clean on-site power — hospitals, corporate campuses, manufacturing plants. This segment provides baseline demand and a diversified customer base outside of the hyper-concentrated data center deals. Growth here is steady but not the inflection driver.

UTILITIES & GRID

Emerging

AEP deal signals utility adoption

The $2.65B AEP deal signals a new channel: utilities deploying Bloom servers to supplement grid capacity in constrained areas. This could become a significant third end market beyond data centers and C&I. If utilities broadly adopt distributed SOFC to bridge grid gaps, the addressable market expands meaningfully beyond the $2.5-5B SOFC TAM.

End-market estimates based on backlog composition and earnings commentary. Bloom does not disclose end-market revenue splits.

Competitive positioning -- power generation landscape

Power Generation Peers -- Revenue, Market Cap, and Technology

Caterpillar (CAT)

Diesel/gas generators | ~33% DC share

$65B rev | $180B mkt cap

GE Vernova (GEV)

Gas turbines, wind, grid | ~24% share

$36B rev | $120B mkt cap

Cummins (CMI)

Diesel/gas gensets | ~15% backup

$34B rev | $50B mkt cap

Generac (GNRC)

Home standby, C&I generators

$4.2B rev | $28B mkt cap

Bloom Energy (BE)

Solid oxide fuel cells | ~40-55% SOFC

$2.0B rev | $68B mkt cap

Peer revenue and market cap from public filings and market data. Market share estimates from industry research.

Oligopoly assessment -- competitive position by market

OLIGOPOLY ✓

Commercial SOFC

~40-55% share | Top 4 = 70-80%

Bloom dominates commercial stationary SOFC. The top 4 players (Bloom, Mitsubishi Heavy Industries, AISIN, Kyocera) hold 70-80% combined, with only 2-3 exceeding 15% share. Competitors sell primarily in Japan for residential/small commercial. No direct commercial-scale Western substitute exists. TAM: $2.5-5B, growing 25-30% CAGR.

FRAGMENTED

DC Behind-the-Meter

~14% share | CAT leads at 33%

In the broader data center behind-the-meter power market, Bloom holds ~14% of tracked projects vs Caterpillar at 33%. At least 5 players have meaningful share including GE Vernova and Siemens. Gas turbines and diesel generators are viable substitutes with different emissions and deployment profiles. TAM: ~$26B.

ROUNDING ERROR

Industrial Power Gen

~2.5% share | $79B market

In the $79B industrial power generation equipment market, Bloom is 1/18th the revenue of GE Vernova and 1/32nd of Caterpillar. Its 2.5% share makes it a niche participant, not a power gen peer. The comp set (GEV, CAT, CMI) is aspirational — Bloom's dominance is technology-specific (SOFC), not market-wide.

Oligopoly assessment from deep dive analysis dated 2026-07-16. Market share estimates from industry research and company filings.

Revenue trajectory -- quarterly trend

Bloom Energy Quarterly Revenue -- Q1 2024 to Q1 2026

Q1 2024

$235M

Q2 2024

$336M

Q3 2024

$330M

Q4 2024

$572M

Q1 2025

$326M

+38.6% YoY

Q2 2025

$401M

+19.5% YoY

Q3 2025

$519M

+57.1% YoY

Q4 2025

$778M

+35.9% YoY

Q1 2026

$751M

+130.4% YoY

Quarterly revenue from Daloopa (company_id: 10820). Q1 2026 +130.4% YoY reflects AI data center power demand inflection.

Key deals and backlog composition

Oracle (2.8 GW commitment). The largest single deal in Bloom's history. Oracle is deploying Bloom Energy Servers to power AI data centers where grid interconnection is unavailable on the required timeline. The 2.8 GW commitment represents years of product deliveries at current manufacturing capacity. Oracle's public endorsement of SOFC technology at this scale validates Bloom's value proposition for tier-1 hyperscalers.

Brookfield ($25B partnership). A joint venture structure where Brookfield provides capital and site development while Bloom supplies Energy Servers. This partnership contributed an estimated 74% of Q4'25 product revenue — a concentration risk flagged by short sellers (Hunterbrook). The JV model accelerates deployments but creates dependency on a single channel partner. Customer diversification beyond Brookfield is a critical watch item.

AEP ($2.65B agreement). American Electric Power represents a new channel: regulated utilities deploying distributed SOFC to supplement grid capacity. If this model scales across the utility sector, it expands Bloom's TAM beyond data centers and opens a third major end market. The AEP deal validates SOFC as grid infrastructure, not just behind-the-meter backup.

Backlog credibility question. Bloom reports a $20B total backlog, but only $493M appears as audited remaining performance obligations (RPO) in SEC filings. The $19.5B gap likely reflects framework agreements, letters of intent, and JV commitments that have not yet met revenue recognition thresholds. Short sellers (Hunterbrook, Crossroads) have flagged this discrepancy as a risk to the bull thesis. Resolution: watch whether quarterly product revenue continues to accelerate as evidence of real backlog conversion.

Data sourced from Daloopa, Bloom Energy SEC filings, earnings transcripts, and Hunterbrook Research. Analysis date: 2026-07-16.