FuelCell Energy, Inc. (FCEL) | The Buildout — AI Infrastructure

——Mkt cap — · 52-wk —–— · YTD — · delayed
Updated Sep 22, 2026Q3 FY2026 reviewed
FuelCell Energy builds molten carbonate fuel cell systems that supply behind-the-meter continuous power for data centers and AI factories.
Fit Energy 380 MW
First data-center order; 30 MW Phase 0 committed with a deposit.
Pipeline ~10 GW
97% data centers, up from 89% in Q2 and over 80% in Q1.
Record cash $737.3M
$658.1M unrestricted; no corporate convertible or high-yield debt.
Q3 gross loss $24.5M
$17M Phase 0 charges; revenue fell 29% y/y.
The Buildout Takeaway
FuelCell Energy has repositioned around AI data-center power, and every forward-looking metric moved sharply higher this year. The reported business has not followed it, and the first data-center order currently costs more to fill than it earns. The question is whether volume and customer elections turn that awarded backlog into contracted, revenue-producing orders — the measure management itself named.
20 analysts·8 Buy10 Hold2 Sell
Coverage is thin — only 6 price estimates, so no target is shown

FY2026 CapEx $10-20M (cut from $20-30M) · Torrington to 100 MW annualized by October 2026 · Fit Energy Phase 0 revenue begins Q4 FY2026 · positive adjusted EBITDA targeted Q4 FY2027 · 500 MW capacity by June 2028 at $200M-$275M, described as fully funded
Important: The Buildout is a data analytics platform. Content is generated by algorithms and AI agents using public filings, earnings transcripts, and market data. This is not personalized investment advice.
Our View

The Verdict

FuelCell Energy manufactures stationary molten carbonate fuel cell systems that generate continuous baseload power, carbon capture, and thermal energy on the customer's own site. Its pitch to data centers and AI factories is speed: a power block that skips multi-year utility interconnection queues and can be brought online as data halls are finished, on the customer's schedule rather than a utility's capital plan. The platform outputs DC power natively, which management aligns with GPU rack architecture, and its high-grade exhaust heat can drive absorption chilling to lower a data center's power usage effectiveness. The company sells power infrastructure into the AI buildout rather than compute or chips — and it competes with turbines, nuclear, and other fuel cell makers for the same load.

Market Cap—
Revenue (TTM)$154M
Revenue Growth+1.1%
EBITDA Margin (TTM)-43.9%
Net Cash$479M
Earnings Beats4 of 7
P/E (TTM)—
EV/EBITDA (TTM)—

What We Like

  • Pipeline grew from more than 1.5 GW cited in Q1 FY2026 to 4 GW in Q2 and roughly 10 GW in Q3, with data centers rising from over 80% to about 97% of it.
  • First commercial data-center order: a capital equipment purchase agreement with Fit Energy covering up to 380 MW across four phases, with 30 MW of Phase 0 committed and an upfront deposit received.
  • Record liquidity: $737.3M of total cash, cash equivalents and restricted cash at July 31, 2026, of which $658.1M was unrestricted, with management stating there is no corporate convertible or high-yield debt.
  • The capacity plan is dated and described as fully funded: 100 MW annualized at Torrington by October 2026, then 500 MW by June 2028 at a cost of $200M to $275M.
  • A service tail behind the hardware: Q2 disclosure set service agreement terms at 15-20 years with a target margin north of 20%, described as generally significantly higher than the initial product sale.

What We’re Watching

  • The first data-center order is loss-making at current scale: $17M of Fit Energy Phase 0 charges in Q3 FY2026, because product costs and manufacturing overhead exceed the contractual pricing at an annualized production rate of about 37 MW.
  • $2.4B of the $3.6B headline backlog is awarded capacity, which the CFO said is "not contracted firm order backlog or a guarantee of future revenue"; Fit Energy's 350 MW Phases 1-3 sit at the customer's sole election with no payment obligation until it elects.
  • Revenue fell 29% year over year and every reported line declined; the CGN Yulchon delivery schedule was not updated on the Q3 call, and the 7.4 MW Groton project is out of service until fiscal 2027.
  • A securities class action emerged after the Q3 call alleging the Fit Energy capacity shortfall was a known trend, with a lead-plaintiff deadline of November 10, 2026.
Bottom Line

The thesis is split rather than simply intact or weakening. Every forward indicator strengthened this year — pipeline, capacity reservations, a first data-center order, and a new awarded-capacity backlog category — while the reported financials weakened, with all four revenue lines down, a $24.5M gross loss, and a first order that costs more to fill than it earns. Management's own framing is that pipeline is a leading indicator, not a result. The open question is whether awarded capacity converts into committed, revenue-producing orders, and whether the company reaches 100 MW of annualized production by October 2026 — the milestone the entire cost-absorption argument rests on.

Next upOctober 2026 is the first test, when management targets 100 MW of annualized production at Torrington. Fit Energy Phase 0 revenue recognition then begins in Q4 FY2026 with the balance in FY2027 — the first hard read on whether volume bends the cost curve.
Last Quarter — Q3 FY2026

Earnings Beat

For its fiscal third quarter ended July 31, 2026, FuelCell Energy reported $33M of revenue, down 29% from $46.7M a year earlier. Gross margin was negative 74.2%, including $17M of Phase 0 charges on the Fit Energy capital equipment purchase agreement — roughly $4M for inventory net realizable value and roughly $13M for losses on firm purchase commitments. The company operated at an annualized production rate of about 37 MW, which the CFO said remains below the volume at which product costs align with market-based pricing. Adjusted EBITDA was negative $36.7M versus negative $16.4M a year earlier, with the Phase 0 charges not added back, and the quarter ended with $737.3M of total cash, cash equivalents and restricted cash.

MetricQ3 FY2026Q2 FY2026Q3 FY2025YoY
Revenue$33M$36M$47M−29.3%
Gross margin-74.2%-36.3%-11.0%-6320bps
EBITDA−$37M−$67M−$86M−56.4%
EPS$-0.63$-1.43$-3.76−83.2%
Awarded capacity backlog$2.4Bn/an/aNew category in Q3 FY2026
Committed backlog$1.3B$1.14Bn/a+~4.1% y/y
I want to be clear that awarded capacity backlog is not contracted firm order backlog or a guarantee of future revenue.— Michael Bishop, CFO, 2026-09-02

Management tone: On the Q2 FY2026 call management's register was confident around commercial engagement — the CEO's line was that the company continues to "emphasize proof over promise." On the Q3 FY2026 call the tone turned operational and execution-focused: the CEO opened with "Our responsibility is straightforward: Execute" and closed with "Opportunity alone does not create value, execution does." Management named the Phase 0 charges and the ~37 MW production rate directly rather than eliding them. The shift from aspirational to operational coincided with a sharply worse printed quarter and the onset of class-action filings.

Management Guidance

On the Q3 FY2026 call, management targeted positive adjusted EBITDA in the fourth quarter of fiscal 2027, assuming "at least 100 megawatts of volume to support that, if not more." It held the 100 MW annualized production milestone at Torrington for October 2026, and reaffirmed 500 MW of annualized capacity by June 2028 at a total cost of $200M to $275M, described as fully funded. Fiscal 2026 capital expenditure guidance was cut to $10M-$20M from $20M-$30M, attributed to equipment delivery timing. Fit Energy Phase 0 revenue is expected to begin in Q4 FY2026 with the balance completed in FY2027. The CFO attached a no-assurance clause covering production rates, backlog conversion, and cost reductions. Product margin guidance of 10% to 20% and service margin guidance north of 20%, both stated in Q2, were not reaffirmed in Q3.

Business Trajectory

Trajectory

Reported revenue is declining while the growth story sits in the pipeline. Trailing-twelve-month revenue was $154.1M with trailing revenue growth of 1.1% year over year, and the latest quarter fell 29%. Part of the decline is mechanical: the Korean GGE repowering completed, taking product revenue to $18M from $26M, and the 7.4 MW Groton project was out of service for the full quarter, taking generation revenue to $8.8M from $12.4M. Gross margin was negative 74.2%, against negative 11.0% a year earlier and negative 36.3% in Q2 FY2026; excluding the $17M of Phase 0 charges, the quarter's gross loss would be roughly $7.5M on $33M, or about negative 22.7% (my arithmetic, approximate). Adjusted EBITDA was negative $36.7M and trailing free cash flow was negative $107.1M. The forward direction depends on a business that has not yet produced revenue.

Revenue & Margin Trajectory
RevenueGross margin$0$20$40$24M$17M$20M$10M$48M$39M$21M$12M$18M$18M$9M$23M$11M$16M$19M$19M$17M$15M$14M$27M$14M$32M$16M$43M$39M$37M$38M$26M$22M$17M$22M$24M$49M$19M$37M$47M$55M$30M$36M$33M-2%-74%Q4'16Q1'17Q2Q3Q4Q1'18Q2Q3Q4Q1'19Q2Q3Q4Q1'20Q2Q3Q4Q1'21Q2Q3Q4Q1'22Q2Q3Q4Q1'23Q2Q3Q4Q1'24Q2Q3Q4Q1'25Q2Q3Q4Q1'26Q2Q3
RevenueGross margin$0$20$40$24M$17M$20M$10M$48M$39M$21M$12M$18M$18M$9M$23M$11M$16M$19M$19M$17M$15M$14M$27M$14M$32M$16M$43M$39M$37M$38M$26M$22M$17M$22M$24M$49M$19M$37M$47M$55M$30M$36M$33M-2%-74%Q4'16Q1'17Q2Q3Q4Q1'18Q2Q3Q4Q1'19Q2Q3Q4Q1'20Q2Q3Q4Q1'21Q2Q3Q4Q1'22Q2Q3Q4Q1'23Q2Q3Q4Q1'24Q2Q3Q4Q1'25Q2Q3Q4Q1'26Q2Q3
Gross margin as reported.
Share Price — 12 Months
$10$20$30$052-wk high $30Sep '25DecMar '26JunSep '26
52-week range $6–$30.
Share Price — 12 Months
$10$20$30$052-wk high $30Sep '25DecMar '26JunSep '26
52-week range $6–$30.
The Numbers

The Model

The model projects FY+1 revenue of $315.0M with EBITDA of negative $24M, a negative 7.55% margin, and FY+2 revenue of $600.0M with EBITDA of $49M, an 8.15% margin. The swing from a negative to a positive EBITDA margin is the whole case: it requires the data-center order to move from charges to recognized revenue and the Torrington production rate to carry the manufacturing overhead. The projection budget carries wide dispersion — FY+1 revenue spans $260M to $365M, a 33% spread, and FY+2 spans $520M to $700M, a 30% spread.

Revenue & EBITDA Projections
REVENUE$158M$315M$600MFY25FY+1 (E)FY+2 (E)EBITDA & MARGIN−$81M−$24M$49M8.2%FY25FY+1 (E)FY+2 (E)
REVENUE$158M$315M$600MFY25FY+1 (E)FY+2 (E)EBITDA & MARGIN−$81M−$24M$49M8.2%FY25FY+1 (E)FY+2 (E)
Solid bars are reported actuals; outlined bars are model projections — not company guidance.
MetricFY2025Next FY (E)Following FY (E)
Revenue$158M$315M$600M
YoY Growth—+99.2%+90.5%
EBITDA−$81M−$24M$49M
EBITDA Margin-51.1%-7.5%8.2%

Projections are the median of 4 independent model runs.

On the Q3 FY2026 call, management targeted positive adjusted EBITDA in the fourth quarter of fiscal 2027, assuming "at least 100 megawatts of volume to support that, if not more." It held the 100 MW annualized production milestone at Torrington for October 2026, and reaffirmed 500 MW of annualized capacity by June 2028 at a total cost of $200M to $275M, described as fully funded. Fiscal 2026 capital expenditure guidance was cut to $10M-$20M from $20M-$30M, attributed to equipment delivery timing. Fit Energy Phase 0 revenue is expected to begin in Q4 FY2026 with the balance completed in FY2027. The CFO attached a no-assurance clause covering production rates, backlog conversion, and cost reductions. Product margin guidance of 10% to 20% and service margin guidance north of 20%, both stated in Q2, were not reaffirmed in Q3.

What Could Go Right — and Wrong

What good looks like
  • Awarded capacity converts into committed backlog — any of Fit Energy's 350 MW Phases 1-3 or the 75 MW Texas reservation moving to definitive agreements.
  • Torrington reaches 100 MW of annualized production in October 2026 and holds it, absorbing overhead and bending the manufacturing cost curve.
  • Fit Energy Phase 0 revenue is recognized from Q4 FY2026 at a margin better than the charges imply, validating the volume-fixes-cost argument.
  • The Q2 product margin target of 10% to 20% and service target north of 20% return to the call language, with the service attach quantified in a data-center context.
  • Rotterdam carbon capture validates under commercial conditions and the ExxonMobil scope widens; the 75 MW Texas counterparty is named with a delivery schedule.
What could go wrong
  • Fit Energy does not elect Phases 1-3, removing $2.4B from the $3.6B headline backlog and leaving a committed book that is still mostly legacy generation.
  • The October 2026 milestone for 100 MW of annualized production slips, breaking the cost-absorption mechanism behind the Q4 FY2027 EBITDA target.
  • Phase 0-type negative gross margin recurs on later phases at higher volume, making the data-center business a revenue story with a negative gross margin attached.
  • Additional equity issuance — share count already rose sharply, and the Q2 answer on capital raises did not rule further raises out.
  • Adverse litigation developments on the known-trend theory, or a competitive default toward gas turbines, nuclear, and gas laterals for data-center load.
What’s Next

Looking Ahead

The next twelve months turn on three verifiable events. Torrington must reach 100 MW of annualized production by October 2026, and Fit Energy Phase 0 revenue must begin in Q4 FY2026 with the balance landing in FY2027. Management's own monitoring metric — the velocity at which awarded capacity converts into committed definitive agreements — is the third. Around those sit a 75 MW Texas capacity reservation awaiting a definitive agreement, Fit Energy's elections on Phases 1-3, no Q3 update on CGN Yulchon deliveries, a Groton upgrade completing in fiscal 2027, and a securities class action with a November 10, 2026 lead-plaintiff deadline. The dated profitability target, positive adjusted EBITDA in Q4 FY2027, depends on most of the above going right.

Catalysts
  • October 2026Torrington hits 100 MW — Annualized production milestone; tests the cost-absorption thesis.
  • Q4 FY2026Fit Energy Phase 0 revenue — First data-center revenue recognition; balance lands in FY2027.
  • H2 FY2026CGN Yulchon deliveries — A Q2 promise toward consistent H2 FY2026 product revenue; no update provided on the Q3 call.
  • November 10, 2026Class-action lead-plaintiff deadline — Deadline to seek lead plaintiff in the Fit Energy securities suit.
  • Q4 FY2027Positive adjusted EBITDA target — Management's dated profitability target; assumes 100 MW+ volume.
  • June 2028500 MW capacity target — Torrington expansion completion.
Numbers

Financials

Annual Summary

MetricFY2024FY2025TTMYoY
Revenue$112M$158M$154M+41.0%
Gross Margin-37.5%-18.9%-32.4%+1,860bps
EBITDA−$122M−$81M−$68M+33.9%
EBITDA Margin-109.1%-51.1%-43.9%+5,799bps
Net Income−$126M−$189M−$176M-49.8%
Free Cash Flow−$212M−$148M−$107M—
Net Cash————

Key Ratios (Trailing)

Valuation
  • P/E TTM—
  • EV/EBITDA TTM—
  • EV/Revenue TTM—
  • Price/FCF TTM—
Profitability
  • Gross Margin (TTM)-32.4%
  • EBITDA Margin (TTM)-43.9%
  • Net Margin (TTM)-114.0%
  • ROIC-17.2%
  • SBC / Revenue6.4%
Reference

The Company

FuelCell Energy is a clean energy technology company and a stationary fuel cell manufacturer, founded in 1969 and headquartered in Danbury, Connecticut, with 22 years of operating experience in the field. It makes proprietary molten carbonate fuel cell systems that deliver large-scale, continuous clean power and emissions management. The platform is fuel-flexible — biofuels, renewable natural gas, or hydrogen-hydrocarbon blends — and provides baseload power, carbon capture, and thermal energy for chilling, heating, and process steam. Two business shapes coexist: a legacy installed-base and generation business, made up of company-owned projects under long-term power purchase agreements, long-term service agreements on customer-owned plants, and module deliveries to utility customers in Korea, which still produces most reported revenue; and an emerging data-center power business, selling behind-the-meter continuous fuel cell power as standardized blocks with capacity reservation agreements and a new awarded-capacity backlog category, which carries the narrative but almost none of the revenue yet.

Manufacturing sits at Torrington, Connecticut — 167,000 square feet, producing cell packages and assembling fuel cell modules, and housing the global service center. Danbury, Connecticut holds headquarters and R&D and is ISO 9001:2015 and ISO 14001:2015 certified. Calgary, Alberta handles solid oxide electrolysis engineering and development, and Taufkirchen, Germany does final module assembly for up to 20 MW per year of sub-megawatt systems serving Europe. The 10-K names three segments — Distributed Generation, Carbon Capture Utilization and Sequestration, and Distributed Hydrogen — but the financial statements report a single segment, "fuel cell power plant production and research," measured on consolidated net loss. The capacity build is funded with equity: the quarter ended July 31, 2026 drew roughly $298M of net equity proceeds.

Business Segments

Distributed Generation
Carbonate fuel cell systems; sold in modular 1.25 MW power blocks scalable to hundreds of megawatts
The cornerstone product line: stationary carbonate fuel cells for utilities, independent power producers, data centers, and behind-the-meter or microgrid applications.
Growth driver: Behind-the-meter data-center baseload power
Carbon Capture, Utilization and Sequestration
Carbonate fuel cell systems with carbon capture capability
Fuel cells that capture carbon while producing power and thermal energy, aimed at data centers, industrial users, and high-CO2 emitters such as oil and gas.
Growth driver: Rotterdam demonstration with ExxonMobil
Distributed Hydrogen
Tri-gen system and solid oxide electrolysis platform, described as under development
The Tri-gen system makes power, hydrogen and water from a single unit; the solid oxide electrolysis platform is engineered at Calgary for industrial hydrogen uses.
Growth driver: First Tri-gen unit at Toyota, Port of Long Beach

Competitive Landscape

The material describes a competitive field in which FuelCell Energy is one option among several. Neighbors in the verified set are contracting gas turbines, nuclear, gas laterals, storage, and third-party data-center capacity — not fuel cells. AEP named Bloom Energy as its early fuel-cell bridging partner, and no neighbor in the verified set named FuelCell Energy on its latest call. The material's criticality assessment concludes that if FuelCell Energy disappeared, AI data centers would readily shift to Bloom Energy or conventional generators, and that its fuel cells are not a chokepoint or a unique solution for AI power. FuelCell Energy's own filing history runs both ways on competition: it previously licensed carbonate fuel cell manufacturing intellectual property exclusively to POSCO Energy in South Korea and broader Asia, and its subsidiary licensed carbonate patents from Fraunhofer IKTS.

  • POSCO Energy
    10-K: FuelCell Energy previously licensed certain carbonate fuel cell manufacturing intellectual property to POSCO Energy on an exclusive basis in South Korea and broader Asian markets.
  • Fraunhofer IKTS
    10-K: FuelCell Energy Solutions GmbH held license rights to 2 U.S. patents and 7 patents outside the U.S. for carbonate fuel cell technology licensed from Fraunhofer IKTS.
  • Bloom Energy
    Named by AEP as its early fuel-cell bridging partner on the neighbor read-through, and listed among inferred competitors; not named as a competitor in FuelCell Energy's own filings.
  • Ballard
    Named on the spider-sourced inferred competitor list; not discussed.
  • Plug Power
    Named on the spider-sourced inferred competitor list; not discussed.
POSCO Energy and Fraunhofer IKTS are documented in the FY2025 10-K; Bloom Energy, Ballard, Caterpillar, Cummins, GE Vernova and Plug Power are spider-sourced inferred competitors, and AEP's naming of Bloom Energy comes from the neighbor read-through, not from FuelCell Energy.

Supply Chain

FuelCell Energy buys balance-of-plant components plus nickel and stainless steel, assembles modules at Torrington, Connecticut, and sells to utilities, data centers and industrial customers. No neighbor in the verified set named it on its latest call.

Supplier
Siemens
Electrical balance of plant, battery storage, microgrid controls, medium-voltage equipment under MOU
Supplier
Balance-of-plant component suppliers (unnamed)
Mechanical and electrical BOP, purchased directly or outsourced to FuelCell Energy designs
Supplier
Nickel and stainless steel
Key inputs in manufacturing, per the 10-K
→
Behind-the-meter continuous baseload power
FCEL
Torrington, CT module assembly — scaling from about 37 MW toward 100 MW, then 500 MW.
→
South Korea (region)
60.1% of Q2 FY2026 revenue
GGE and CGN module and service programs
Fit Energy
380 MW CEPA
30 MW Phase 0 committed; 350 MW at the customer's election
ExxonMobil / Esso Nederland
$11.6M purchase order
Rotterdam carbon capture demonstration modules
U.S. Navy
7.4 MW
Groton microgrid, out of service for upgrade
Major co-location operator (unnamed)
75 MW
Texas capacity reservation, closed after quarter-end

Analysis updated Sep 22, 2026, reviewing Q3 FY2026. Prices delayed. Built with The Buildout’s published methodology. Not investment advice. No positions held. © The Buildout 2026.

More on FCEL: Earnings recap