FuelCell Energy's Q3 Put a Price Tag on Its Data-Center Bet

FuelCell Energy reported its fiscal third quarter on September 2, and the stock took the news the way the numbers deserved — down sharply again, roughly 25% lower over the past week after a year that carried it from about $4 to a $38 high. The selloff is the least interesting part. What matters is a line buried in the release: at the company's current production rate, the cost of building the fuel cell that powers its flagship data-center deals exceeds the price the marquee contract actually pays.

Here is the quarter. Revenue fell 29% year over year to $33 million, short of what analysts had expected, and the loss per share of 64 cents came on a net loss of about $45 million. The pain was concentrated in gross margin, which swung to a $24.5 million gross loss, and the adjusted EBITDA loss widened to roughly $37 million. FuelCell attributes the damage to inventory and purchase-commitment charges tied to Phase 0 of its Fit Energy agreement, and it concedes that costs at the current ~37 MW annualized production rate run above what that contract prices in. In plain terms: today, each unit the company is selling into its biggest new deal loses money.

None of that is the whole company. This is still a business with real, reportable assets on the other side. Cash hit a record $737 million — a pile that represents about two-thirds of the current market value. Committed backlog is $1.3 billion and, adding uncontracted awarded capacity, $3.6 billion, much of it data-center load. After the quarter it signed a capacity reservation for a 75 MW data-center project in Texas, and it renewed its big agreement with Fit Energy for up to 380 MW. Management repeated specific ramp targets: 100 MW of annualized production by October, 500 MW of capacity at its Torrington plant by June 2028, and positive adjusted EBITDA by fiscal Q4 2027.

This is where I want to be honest about what the quarter is and is not. My usual setup is the beaten-down name where the market has given up while the numbers quietly improve — tape pain without business pain. FuelCell is the opposite shape. The stock ran up on the data-center narrative long before the economics arrived, and in the quarter just reported the numbers moved the wrong way: revenue down, margins worse, the flagship deal unprofitable at current production. That is operating pain, not just ugly sentiment.

So the rerating now depends on one checkable thing, and it is not the backlog or the cash. It is whether cost per megawatt falls below contract price as production scales from ~37 MW toward 100 MW and then 500 MW. That is the bridge between the story and the free cash flow — and free cash flow is still deeply negative here, which is why I am not attaching a price target. I want to see the margin converge first.

Watch two tripwires. The first is margin: if quarterly gross losses keep widening as volume climbs, the ramp is not making each unit cheaper, and the break-even-by-2027 target quietly slips. The second is dilution. That record cash balance did not come from earnings; it was topped up by selling stock, including a July offering and shares sold into the market at an average near $13. The cushion funds the build-out, but it is replenished with shareholders' capital, not retained profits. Every raise buys time but also raises the per-share bar.

I can be wrong again — a fuel-cell maker scaling to 500 MW could surprise in either direction. But the honest reading of Q3 is that the market bought the promise before the profit, and the quarter just made the cost of that promise visible. The thesis is not dead. It is now contingent on a specific, checkable number: whether the gross margin closes the gap as production triples. If it does, the story gets a lot harder to dismiss. If it does not, the cash keeps the company alive long after the narrative runs out of patience. Watch the margin, not the headlines.