FuelCell Energy's New Bull Case Depends on Turning Optional Backlog Into Real Orders

FuelCell Energy has become a clean way for investors to express one of AI's most urgent constraints: getting large amounts of power to data centers faster than the grid can deliver it. Oppenheimer added fuel to that thesis on September 29, initiating FCEL at Outperform with a $24 price target.

The rating is understandable. FuelCell has a roughly 10 GW sales pipeline, its first capacity reservation agreement with a major data-center operator and plans to expand Torrington manufacturing capacity to 500 MW annually by June 2028. But the headline backlog requires much more scrutiny than the bullish narrative usually gives it.

The $3.65 billion backlog is not all the same

At July 31, FuelCell reported $1.296 billion of committed backlog plus $2.350 billion of awarded capacity backlog. The second bucket relates to optional future phases of the Fit Energy agreement covering as much as 350 MW.

FuelCell's own disclosure is unusually clear: awarded capacity backlog is not contracted backlog, firm order backlog or a guarantee of future revenue. Fit Energy may elect whether to proceed, and site selection, permitting, financing and construction can still stand between an award and an operating project. Investors who add the two categories together without discounting the optional portion are overstating what FuelCell has actually won.

The manufacturing ramp can create operating leverage, or expensive idle capacity

FuelCell is targeting a 100 MW annualized production rate in October 2026 and 500 MW of annual capacity by June 2028. That is exactly the kind of scale required if large data-center projects convert.

It also creates a classic industrial risk. Factories earn attractive returns when orders fill them. If optional projects slip, new equipment and fixed costs arrive before revenue. Management is targeting positive adjusted EBITDA in fiscal Q4 2027, but explicitly conditions that goal on awarded backlog converting, customer schedules and cost reductions.

The latest quarter does not yet look like a breakout business

Q3 revenue was $33.0 million, down 29% year over year, and gross loss widened to $24.5 million. The balance sheet provides time: cash, cash equivalents and restricted cash totaled $737.3 million.

That cash bridge is important because the company is scaling ahead of the revenue it expects. The investment case therefore depends less on whether AI needs power, which is increasingly obvious, and more on whether FuelCell can convert interest into deposits, binding orders and profitable production before the expansion consumes too much capital.

GMR view: the power thesis is stronger than the stock thesis

We agree with the core idea behind the bullish initiation: behind-the-meter generation has real strategic value when grid interconnections take years. FuelCell has credible technology, a real data-center reservation and enough liquidity to attempt the manufacturing ramp.

But we think FCEL should be valued on committed orders and conversion evidence, not on the full awarded-capacity headline. The market is already paying for a future in which optional megawatts become factories running near capacity. Until gross margins turn positive and awarded backlog repeatedly converts into firm business, FuelCell remains a high-upside infrastructure option rather than a proven AI power compounder.