U.S. · ALPHAGBM RESEARCH

Bloom Energy: Growth and execution in the AI power cycle

2026-09-10 · 18 min · AlphaGBM
BE

Open the interactive research edition →

Bloom Energy · Growth and execution in the AI power cycle

NYSE · United States · BE

From time-to-power to orders, earnings and cash flow. The research case for SOFC extends beyond power shortages: can the company turn its technology into sustainable operating returns?

2026-08-24 (U.S. Eastern) · 2026-08-25

AGBM-TH-2026-0825-01

Operating improvement is visible in the financial results. The next test is whether it can withstand project delivery, customer concentration and capital-spending pressures.

Revenue scale is expanding

Product earnings dominate

Execution conditions still matter

MetricValueUnitBasis
Q2 2026 revenue10.65USD 100mYear-on-year +165.5%
GAAP gross margin33.4%26.7% a year earlier
Operating cash flow2.26USD 100mQ2 alone, not the full year
Full-year revenue guidance39–42USD 100mCompany guidance dated July 28; not realized revenue

Delivery speed creates an entry point

Both Oracle and Nebius announcements emphasize time-to-power. Value comes from meeting specific projects' power needs sooner, not from assuming every data center will use the same generation technology.

Growth and profitability are improving together

Q2 revenue and GAAP operating profit both increased. Product revenue remains dominant, however; non-product revenue should not all be treated as stable service subscriptions.

A framework is still several steps away from cash

Oracle, Brookfield, AEP and Nebius represent different contractual stages. Separating money from capacity, and plans from deliveries, makes the evidence behind future revenue clearer.

Business and industry: why time-to-power matters

BUSINESS & DEMAND

AI data-center construction and power-infrastructure expansion do not always move together. Bloom's entry point is to provide usable power sooner at suitable sites.

Bloom Energy's core Energy Server uses solid oxide fuel cells (SOFC). Electrochemical reactions convert fuel's chemical energy into electricity in a distributed generation system that can be installed near demand. Research should consider equipment, long-term maintenance and project delivery rather than treating the business simply as an AI-themed stock.

The demand thesis is not that the grid will be replaced. Customers weigh capacity, reliability, delivery dates, permitting and costs. Earlier power availability can affect when data-center racks enter service, but fuel access, site conditions, acceptance and lifetime power costs also determine whether a project is viable.

From fuel to compute: value and constraints

Fuel and site

Suitable fuels such as natural gas, supply reliability, permitting and connection requirements

SOFC system

Electrochemical generation; manufacturing quality, efficiency and stack maintenance

Data-center power

Grid-connected or off-grid configuration, redundancy, power quality and acceptance

Compute services

Rack commissioning, customer workloads, service contracts and cash collection

[8] US DOE — fuel-cell technology comparison · reference

[3] Bloom Energy / Oracle partnership · 2026-04-13

[5] Nebius / Bloom Energy first deployment · 2026-05-20

Demand context: data-center electricity use

Historical estimate / Scenario forecast · TWh / year · 2026-08-24T00:00:00Z

2024 estimate2030 base case
IEA base case415945

IEA's 2025 Energy and AI base case. The 2030 figure is a forecast, not realized demand, and does not directly imply Bloom orders.

[7] IEA Energy and AI — energy demand · 2025-04-10

The IEA base case supports the background of rising data-center electricity demand, but industry growth is not the same as company market share. Customer choice, financing, alternatives and local policy determine whether demand becomes orders. Distributed generation is one part of the supply mix, alongside grid expansion, gas generation, storage and load management.

Equipment sales

Hardware delivery and project scale influence the timing of revenue; manufacturing efficiency and product mix affect gross margin.

Installation and maintenance

Installation, services and electricity sales are distinct businesses. Recurring revenue depends on contracts, maintenance costs and customer retention.

Limits to commercial value

Technical suitability does not ensure attractive economics. Gas prices, financing, taxes, site conditions and redundancy can change customer choices.

Operating performance: does growth create leverage?

OPERATING PERFORMANCE

Start with consecutive quarters, then compare earnings with cash flow. One high-growth quarter cannot establish the overall quality of earnings.

Five consecutive quarters: revenue, profit and margin

Quarterly revenue

Company disclosure · USD million · 2026-08-24T00:00:00Z

25Q225Q325Q426Q126Q2
Revenue401.242519.048777.683751.0541065.365

Consecutive quarters explicitly reported in two company releases; no quarterly observations are invented from annual guidance.

[1] Bloom Energy Q2 2026 results · 2026-07-28

[2] Bloom Energy FY2025 results · 2026-02-05

GAAP operating profit

Company disclosure · USD million · 2026-08-24T00:00:00Z

25Q225Q325Q426Q126Q2
Operating profit-3.5037.84687.52972.19182.237

GAAP operating profit, including losses; not mixed with non-GAAP profit.

[1] Bloom Energy Q2 2026 results · 2026-07-28

[2] Bloom Energy FY2025 results · 2026-02-05

GAAP gross margin

Company disclosure · % · 2026-08-24T00:00:00Z

25Q225Q325Q426Q126Q2
Gross margin26.729.230.83033.4

Percentages are as reported. Changes between rounded percentages are not exact basis-point changes calculated from underlying amounts.

[1] Bloom Energy Q2 2026 results · 2026-07-28

[2] Bloom Energy FY2025 results · 2026-02-05

Revenue does not rise in a straight line every quarter: Q1 2026 fell from Q4 2025 before Q2 increased significantly. Delivery and acceptance timing affect revenue recognition, so the trend alone does not establish that every quarter will accelerate.

Q2 GAAP gross profit was $355.6 million, operating expenses $173.3 million and operating profit $182.2 million. Gross-profit growth exceeded expense growth, indicating operating leverage. Its durability still depends on delivery scale, product margins and expense control working together.

Financial comparison: consistent definitions and periods

Metric2025 Q22026 Q12026 Q2
Revenue · USD million401.242751.0541,065.365
GAAP gross margin26.7%30.0%33.4%
GAAP operating profit · USD million−3.50372.190182.237
Net income attributable to common stockholders · USD million−42.61970.653196.290
GAAP diluted EPS · USD−0.180.230.62
Operating cash flow · USD million−213.11173.610226.432

Net income uses the common-stockholder measure and is not interchangeable with net income including non-controlling interests.

[1] Bloom Energy Q2 2026 results · 2026-07-28

Revenue mix: more than product sales

Company disclosure · USD million · Q2 2026 · 2026-08-24T00:00:00Z

ProductInstallationServiceElectricity
Revenue by segment935.41350.97869.0239.951

Products represent approximately 87.8% of revenue and services 6.5%. Installation and electricity should not be combined with services and relabeled recurring maintenance revenue.

[1] Bloom Energy Q2 2026 results · 2026-07-28

Revenue still comes mainly from product deliveries. Services can add continuity, but their scale and cost structure differ. A claim that the model is shifting toward long-term services requires monitoring service revenue, service margins and maintenance obligations, not merely the non-product share of revenue.

From accounting earnings to cash: three comparison quarters

Company disclosure · USD million · 2026-08-24T00:00:00Z

2025 Q22026 Q12026 Q2
Operating cash flow-213.11173.61226.432
Cash purchases of property and equipment7.24526.18251.641

These are the comparison periods supplied in the announcements, not consecutive quarters. Property-and-equipment spending is shown as a positive outflow amount and is not added to operating cash flow.

[1] Bloom Energy Q2 2026 results · 2026-07-28

Cash improvement matters more than a first-positive claim

Q2 operating cash flow less property-and-equipment purchases was approximately $174.8 million. This simplified free-cash-flow calculation excludes other investing cash flows and does not establish a first-ever positive result. Receivables and inventory still tie up cash; follow-up should assess working-capital needs as the business grows.

Liquidity

Cash and cash equivalents were approximately $2.6669 billion at June-end; total current assets were $4.5901 billion. Cash is not synonymous with fully distributable funds.

Share-count definitions

Period-end shares outstanding were approximately 293.4 million, versus quarterly weighted-average diluted shares of 323.3 million. These serve different purposes and are not interchangeable for historical market capitalization or scenario share prices.

Customers and projects: from scale to revenue evidence

CUSTOMERS & DELIVERY

Frameworks, purchase arrangements, project plans and delivered capacity belong to different evidence categories. They should not be combined into an apparently certain order total.

Four relationships, four execution paths

Oracle

1.2 GW · Initial capacity being deployed

The April 13 announcement states that the master services agreement supports up to 2.8 GW, with an initial 1.2 GW being deployed. The initial capacity is included in the framework and must not be counted twice.

[3] Bloom Energy / Oracle partnership · 2026-04-13

Nebius

328 MW · Planned capacity of the initial project

The May 20 announcement expects the first 328 MW project to enter operation in 2026. This is a project plan and expected timing, not proof that all capacity has already been accepted.

[5] Nebius / Bloom Energy first deployment · 2026-05-20

AEP

100 MW · Disclosed initial order

The 2024 impact report describes an agreement for up to 1 GW and an initial 100 MW order. Subsequent purchases should not be treated as already fully ordered.

[6] Bloom Energy 2024 impact report · 2025

Brookfield

$25 B · Power-project financing framework

The June 30 announcement expands the financing framework from $5 billion to $25 billion. This is not recognized Bloom revenue and cannot be converted into GW without engineering-cost assumptions.

[4] Bloom Energy / Brookfield financing framework · 2026-06-30

These partnerships span different customer structures: hyperscaler self-builds, utility procurement, AI-cloud projects and infrastructure financing. Broader customer types can reduce reliance on one procurement mechanism, but customer concentration, delivery concentration and shared financing constraints may remain.

Track scale by stage: Oracle's 2.8 GW framework includes the initial 1.2 GW, while Brookfield is a financing amount. Executable delivery schedules, acceptance terms and collection progress are more informative about revenue than framework totals.

How a partnership reaches the financial statements

Cooperation framework

Defines the scope; may include conditions and subsequent options

Specific order

Defines specifications, price, quantity, payment and performance obligations

Delivery and acceptance

Manufacturing, installation, connection and customer acceptance

Revenue and cash

Recognize revenue under accounting requirements, then verify contractual cash collection

[3] Bloom Energy / Oracle partnership · 2026-04-13

[5] Nebius / Bloom Energy first deployment · 2026-05-20

[1] Bloom Energy Q2 2026 results · 2026-07-28

Research judgment

Revenue visibility improves through executable schedules, payment arrangements, delivery acceptance and continuing disclosures, not merely customer names or framework size. Track these states project by project to avoid counting the same project across multiple announcements.

Competition and barriers: technology and economics

COMPETITION & MOAT

SOFC has characteristics suited to stationary generation, but suitability for data centers does not mean an absence of alternatives.

Bloom, Plug Power and FuelCell Energy all involve fuel cells, but their technology, business mix and project stages differ. First identify what the customer buys: stationary power, backup generation, or hydrogen production and logistics. Then compare profits, cash consumption and delivery capability over consistent periods.

Fuel-cell technologies: advantages and trade-offs

TechnologyTypical characteristicsMain constraintsResearch question
SOFC · Solid oxideHigh-temperature operation and fuel flexibility; suitable for stationary generation and combined heat and powerMaterials durability, hot components, start/stop operation and maintenanceCan sustained output cover maintenance and replacement costs?
PEM · Proton exchange membraneLower-temperature operation and faster startup, with distributed-power and transport applicationsCatalyst cost, fuel purity and hydrogen supplyDo the customer's load profile and fuel conditions favor this technology?
MCFC · Molten carbonateHigh-temperature operation and fuel flexibility for stationary generation and combined heat and powerCorrosion, component durability and startup timeHow does the specific project balance efficiency, maintenance and integration?

The DOE comparison explains technology categories; it does not rank listed companies' products or investment value.

[8] US DOE — fuel-cell technology comparison · reference

Manufacturing and reliability

Ceramics, manufacturing consistency and operating history create accumulated know-how. Evidence should include failure rates, service life, maintenance costs and repeatable delivery, not patent counts alone.

Integration with customer systems

Power solutions involve redundancy, site design, maintenance and responsibility allocation, potentially creating switching costs. Renewals and follow-on orders must still demonstrate the advantage.

Scale and financial capacity

Procurement, manufacturing and service scale may improve costs. Capacity expansion nevertheless consumes funding; improved earnings do not establish permanent independence from capital markets.

Relevant alternatives also include grid connections, gas equipment and other power combinations. Customers care about reliable capacity available by a deadline, lifetime cost, and who bears construction and operating risks. A technical advantage is not automatically a commercial advantage in every setting.

Transaction disclosures: an event is not an answer

PUBLIC DISCLOSURE

Congressional transaction filings can generate research leads, but do not prove motives, informational advantages or future returns.

The Pelosi-related public filing lists owner code SP. Purchases of BE shares and call options occurred on July 24 and July 28. The filing identifies instruments, quantities, dates and amount bands; a band's upper bound is not the actual transaction amount.

BE disclosures by transaction date and instrument

DateReported instrument and quantityDisclosed amount bandWhat the filing establishes
2026-07-2410,000 common shares$1,000,001–$5,000,000Reported purchase quantity and amount band
2026-07-24100 call option contracts$1,000,001–$5,000,000$100 strike; expiration 2027-06-17
2026-07-285,000 common shares$500,001–$1,000,000A second filing entry, separate from July 24
2026-07-28100 call option contracts$500,001–$1,000,000Same strike and expiration; a separate reported transaction

House PTR #20035143. Quantities and amount bands come from the filing and do not establish execution prices or option profit and loss.

[9] 美国众议院 PTR #20035143 · 2026-08-21

How to use this lead

It can prompt questions about industry interest, but cannot replace research into orders, cash flow and valuation. First-ever, smart-money and motive claims need independent evidence beyond a single filing.

Valuation: make the conditions explicit

VALUATION & SENSITIVITY

Rather than a price target, show the scenario equity value implied by specified revenue and multiple assumptions.

The company's July 28 guidance for 2026 revenue was $3.9–4.2 billion, with a $4.05 billion midpoint. Against actual 2025 revenue of approximately $2.0240 billion, this implies growth of about 92.7–107.5%, or 100.1% at the midpoint. Guidance is management's expectation, not completed revenue.

Company guidance: keep the measures separate

MeasureFull-year 2026 guidanceResearch use
Revenue$3.9–4.2 billionReference range for revenue scenarios
Non-GAAP gross marginApproximately 34%Compare operating structure and earnings quality
Non-GAAP operating profit$800–900 millionNot net income or cash flow
Non-GAAP EPS$2.55–2.85Retain the diluted-share and non-GAAP definitions

Information retains the original report cutoff; later updates are excluded.

[1] Bloom Energy Q2 2026 results · 2026-07-28

Revenue × P/S: conditional equity-value sensitivity

Conditional calculation, not a price target or return promise.

Revenue × P/S; revenue and equity value in USD 100 million.

Revenue121928
39.0468.0741.01092.0
40.5486.0769.51134.0
42.0504.0798.01176.0

Revenue and equity value are shown in USD 100 million. The original 12×/19×/28× multiples are scenario assumptions, not verified fair-value multiples. Results are conditional equity market values, not enterprise values or per-share targets.

[1] Bloom Energy Q2 2026 results · 2026-07-28

Operating conditions disappoint

Delayed delivery, product-margin pressure or greater working-capital consumption could reduce both revenue and the multiple investors are willing to pay.

Existing projects deliver as planned

Meeting guidance still requires delivery and cash collection. Reaching the revenue midpoint does not automatically justify any particular valuation multiple.

Conditions improve further

Additional executable projects, stable cash generation and stronger customer continuity may affect valuation. That requires new evidence, not simply a higher multiple.

P/S calculated using full-year revenue guidance is forward-looking, not trailing twelve months. A per-share conversion also needs the share count at the valuation date and potential dilution, rather than an unadjusted quarterly weighted-average share count.

P/S comparability depends on revenue recognition, margins, capital expenditure and leverage. Comparisons with REITs or other infrastructure companies should first explain those differences; identical multiples do not imply identical cash returns or risk.

Counterevidence and follow-up

RISKS & CHECKPOINTS

Research quality depends not only on supporting evidence, but also on identifying what would weaken the judgment.

Delivery and customer concentration

Schedule changes, delayed acceptance or volatile revenue from a major customer

Verify deliveries, customer advances, revenue and receivables by customer and project.

Profitability and maintenance costs

Product-margin improvement fades, or service obligations and maintenance costs exceed expectations

Track segment margins, warranty and maintenance expenses, and delivery mix.

Fuel, policy and alternatives

Gas prices, taxes, permits or alternative power solutions change customer economics

Compare lifetime project costs, not equipment delivery speed alone.

Valuation and funding

Revenue growth fails to support high multiples while expansion and project funding needs rise

Recheck revenue, cash, debt and share-count changes together and recalculate scenarios.

Follow-up checklist: what to examine after an event

Triggering eventEvidence to updateEffect on the judgment
Next quarterly resultsRevenue, segment gross margins, operating cash flow, receivables and inventoriesTest whether operating leverage persists rather than focusing on one growth rate
Major customer project updatesSpecific capacity status, acceptance, payments and commissioning plansIncrease or reduce visibility into revenue conversion
New financing or equity changesIssuance, conversions, debt maturities and the corresponding share-count basisRecalculate per-share measures and liquidity instead of reusing stale inputs
Policy or power-cost changesAffected jurisdictions, project impacts and allocation of costs to customersTest whether technical suitability remains economically attractive

This is a research checklist, not an enabled automatic monitor, trading system or notification service.

[1] Bloom Energy Q2 2026 results · 2026-07-28

[3] Bloom Energy / Oracle partnership · 2026-04-13

[5] Nebius / Bloom Energy first deployment · 2026-05-20

Source and revision notes
OriginalSectionRevision
Cover and investment themes · P1–2Opening page and executive summaryPreserve the three research themes; remove the unverified first-position claim and unsupported certainty.
Company overview · P3Business and industry / Competition and barriersAdd technical constraints; correct claims of a sole solution and one-dimensional efficiency rankings.
Financial analysis · P3–5Operating performanceRetain five consecutive quarters, revenue mix, cash comparisons and tables; distinguish GAAP measures and share-count definitions.
AI power and customers · P5–6Business and industry / Customers and projectsAdd IEA and customer-announcement evidence; correct Oracle's framework to 2.8 GW and remove incomparable totals.
Transaction disclosures · P2Transaction disclosuresSeparate July 24 and July 28, retain disclosed ranges, and remove inferred motives and exact amounts.
Competitive landscape · P7Competition and barriersRetain technology and business comparisons without ranking profits from mismatched reporting periods.
Valuation analysis · P7–8Valuation and sensitivityRetain the original multiples as explicit assumptions; correct arithmetic and forward-period labels without generating unverified per-share targets.
Risks and disclosures · P9Counterevidence and follow-up / SourcesPreserve material risks, consolidate repeated disclosures and add concrete research checkpoints.

This report is for research reference, not personalized investment advice, trading instructions or a return guarantee. Information retains the original report cutoff. Forecasts, scenarios and company plans are not realized outcomes. Identified errors have been corrected; references that lack independent confirmation remain explicitly qualified. This does not claim that all external verification is complete. English translation prepared with AI assistance; refer to the Chinese edition for the original analysis.

Continue this validation in Alpha Agent with data, counterevidence, and invalidation conditions. Explore AlphaGBM →