Summary

  • FuelCell Energy’s June agreement with Fit Energy USA LP gives the buyer a 30MW Phase 0 with effective payment obligations and an initial deposit received; it does not say that 30MW has been delivered, accepted or is earning revenue.
  • The same agreement gives Fit sole election rights over three later phases of 100MW, 125MW and 125MW. Those 350MW become payment obligations only if Fit elects them on time and the next project agreements are reached.
  • A post-quarter 75MW Texas capacity reservation gives an unnamed major data-centre operator priority access to manufacturing capacity while definitive project documents are being finalised. It is neither a disclosed operating project nor evidence that the reservation is an additional contracted build.

The phrase “up to 380MW” can make one commercial relationship look like one completed asset. FuelCell’s filed documents instead contain several receipts with different owners, conditions and clocks. The distinction is not a semantic downgrade of the company’s opportunity. It is the only way to see what has actually been purchased, what a counterparty may still choose, and what production capacity has merely been held open.

A 30MW first receipt, not a 380MW completion receipt

FuelCell’s Form 8-K says its capacity purchase and electrification agreement, or CEPA, with Fit Energy USA LP covers a 30MW Phase 0 and later potential phases. The July 31 Form 10-Q is more specific about the first state: Fit’s payment obligations for Phase 0 became effective on June 24 and FuelCell received an initial deposit. The company expected delivery of the Phase 0 modules in the fourth quarter of fiscal 2026.

Those are meaningful facts. An effective payment obligation and a received deposit are different from a non-binding announcement. But delivery was still expected, not reported as completed. The reviewed filings do not establish a completed installation, commissioning, customer acceptance, energy delivery or Phase 0 revenue. The proper current label is an effective initial commercial obligation with an expected delivery date.

That boundary protects both readers and the company’s evidence. A manufacturer can have a valid order before the modules are built, shipped, installed and accepted. Turning the valid order into operating data-centre power, however, consumes several unreported receipts in advance.

The 350MW is Fit’s election surface

The warrant agreement lays out the next three phases: 100MW, 125MW and 125MW. Together they total 350MW. They are not simply the unsupplied balance of one 380MW firm purchase. Fit has the sole right to elect each phase, and its payment obligation for an elected phase becomes effective only on a timely election under the agreement.

The same documents put further project work between an election and physical delivery. As sites are identified, the parties are expected to enter prescribed project commissioning agreements and long-term service agreements, with FuelCell describing the latter as expected to run 15 to 20 years. Site identity, the final project documents, permits, financing, construction responsibility, ownership and end-customer arrangements remain separate questions unless a later filing supplies them.

This is a valuable option architecture for both sides. Fit can preserve access to potential capacity while matching election to its projects; FuelCell can present a large addressable relationship without treating every later phase as a current manufacturing obligation. The analytical error is to replace the word “option” with “backlog” before the election and associated contracts exist.

A separate 75MW reservation is a queue position

After the quarter, FuelCell announced a capacity reservation agreement for up to 75MW with an unnamed major data-centre operator in Texas. It says the operator made an upfront reservation payment and gained priority access to FuelCell manufacturing capacity while the parties finalise definitive project agreements.

That payment makes the reservation more substantial than an expression of interest. It does not collapse the reservation into a completed purchase agreement. The company did not name the operator or identify a facility, a completed site, a final equipment configuration, an energisation date, a construction notice or a revenue start. “Priority access” describes a production-queue right. It is not evidence that 75MW is operating, installed, financed or even a firm order on the same terms as Phase 0.

Readers should also resist adding 75MW mechanically to the Fit relationship. The disclosure identifies a different counterparty and a different document. It may become a project; it may change; it may lapse or evolve under its definitive agreements. Until then, it is a paid reservation with an explicit contracting condition.

The cost receipt exposes the first phase’s operating question

The July-quarter 10-Q adds a harder, more useful piece of evidence. It reports approximately US$4 million of inventory net-realizable-value charges and about US$13 million of firm purchase commitments in product cost connected to the Phase 0 CEPA. It also says product costs and manufacturing overhead exceeded CEPA pricing at the disclosed annual Torrington production rate of 37.1MW, below the volume the company expected would align its cost base and pricing.

This is not proof that the overall arrangement is uneconomic, nor does it price a 350MW election that has not occurred. It is a receipt about the economics of the current initial phase at a stated production rate. It tells a reader to distinguish a capacity headline from the margin, volume and production assumptions required to make the relationship work commercially.

Keep three ledgers, then a fourth

The Phase 0 ledger asks for manufacturing, delivery, commissioning, acceptance and revenue evidence. The option ledger asks whether Fit elects 100MW, 125MW or 125MW phases on time and whether site-specific commissioning and service agreements are signed. The reservation ledger asks whether the Texas counterparty signs definitive documents and converts its queue position into an identified project. The economics ledger asks whether volumes, costs, overhead absorption and price move toward the company’s stated alignment.

Each ledger can progress without proving the others. A 30MW delivery would not make 350MW elected. A 75MW reservation would not cure Phase 0 margin pressure. A future service agreement would not itself prove energisation or collected revenue. The headline becomes more informative, not less, once its separate receipts remain separate.

Sources