AI data centers are beginning to pay a premium for something more valuable than cheap electricity: time.
A grid connection can take years. So can delivery of a large gas turbine. A fuel-cell installation can begin producing power in months.
That is turning fuel cells into an escape route from the power bottleneck and the escape route is beginning to tighten.
Tessara currently reads Fuel Cell Power as Tight at 79. Tight, but not yet Critical. The score is being driven primarily by unusually strong commercial demand, rather than evidence that customers cannot obtain fuel-cell hardware.
The Fuel Cell Power constraint is currently Tight levels (Source: Tessara)
That distinction is important because fuel cells are not yet a classic shortage. They are becoming valuable because other parts of the power stack are too slow.
That is the constraint we are looking at this week.
What is a fuel cell?
A fuel cell is essentially a power plant with no combustion step.
Instead of burning fuel to make heat and then converting that heat into electricity, a fuel cell converts the chemistry directly into electrical current.
Bloom Energy uses solid oxide fuel cells. Air runs across one side of a ceramic electrolyte and natural gas across the other. Oxygen ions move through the ceramic, while electrons are forced around an external circuit. That detour produces electricity.
How Solid Oxide Fuel Cell Works (Source: Bloom Energy)
A single cell produces roughly a volt, so cells are stacked into modules and modules into larger systems. That modularity is part of the appeal. Capacity can be added by repeating the same unit rather than building one enormous machine.
Two features matter for data centers:
First, there is no flame. Fuel cells still emit carbon dioxide when running on natural gas, but avoiding combustion changes the local emissions profile and can simplify parts of the permitting process.
Second, the output is direct current. AI racks ultimately consume DC power, so a fuel cell system can reduce some of the conversion steps between generation and the rack.
But the main advantage is that fuel cells solve for time.
They can sit behind the meter, use existing natural-gas infrastructure and begin producing power far sooner than a new grid connection or large turbine deployment.
What Is Driving the Constraint
What is holding the score up is a demand step-up that has already happened, led overwhelmingly by Bloom Energy.
Bloom recently crossed $1 billion of quarterly revenue for the first time, with both sales and margins moving sharply higher. That matters more than another announced gigawatt-scale campus because this demand has already moved through the pipeline and onto the income statement.
Bloom Energy Revenue data series on Tessara
Behind it sits another layer of potential demand.
Oracle is moving ahead with gigawatt-scale Bloom deployments. Brookfield has expanded the financing available for similar projects. FuelCell Energy says its data-center opportunity set has grown into a multi-gigawatt pipeline.
Those signals mean different things:
Bloom shows demand that has already converted.
FuelCell Energy shows demand that could convert next.
Oracle and Brookfield show how much larger the market could become if project pipelines keep moving forward.
That is enough to keep the constraint elevated. But it is not the same as evidence of customers waiting for hardware that manufacturers cannot supply.
The supply base is already responding.
Bloom is moving annual manufacturing capacity from roughly 1 GW to 2 GW by the end of 2026, while management says the same footprint can eventually support around 5 GW.
FuelCell Energy plans to expand its Torrington facility from roughly 100 MW toward as much as 500 MW as its data-center pipeline grows.
This is the key difference between fuel cells and many of the bottlenecks they are helping data centers bypass. A transmission line or large turbine supply chain can take years to expand. Fuel cell systems are modular and more cells, stacks and cabinets can be produced without waiting for an entirely new class of generation asset to be built.
So far, manufacturing has responded quickly enough that strong demand has not become outright hardware scarcity.
That is why the constraint is Tight, not Critical.
Who Captures the Economics?
The market is often discussed as if every fuel cell company benefits equally from tighter power availability. They do not.
Three things matter:
Exposure to the data center demand driving the constraint
Whether that demand is already becoming revenue
Whether the underlying technology fits the projects being built today.
Fuel Cell Power Constraint’s Beneficiaries (Source: Tessara)
Bloom Energy: converting the demand now
Bloom has the clearest exposure to the speed-to-power trade.Its solid oxide systems run primarily on natural gas, making them a practical fit for campuses that can access pipeline gas but cannot wait years for grid capacity. Large deployments are already showing up in revenue and margins.
Bloom has therefore moved from being positioned for the bottleneck to actually monetising it.
The key signal now is whether demand keeps outrunning the manufacturing ramp. If capacity expands and delivery times still begin to stretch, Fuel Cell Power would be moving from demand-led tightness toward a genuine hardware constraint.FuelCell Energy: the next conversion test
FuelCell Energy uses molten carbonate fuel cells, another gas-fed technology that can serve large stationary projects.Its data-center pipeline has grown into multiple gigawatts, and it is expanding manufacturing toward that opportunity. But the economics have not followed yet.
That makes FuelCell Energy useful as a forward indicator. If its pipeline starts converting into signed deployments and revenue, the demand signal broadens from a Bloom story into a more general fuel cell market signal.Plug Power: a different market
Plug Power primarily uses PEM fuel cells, which depend on hydrogen.
That leaves it much less directly exposed to the current data center trade, where the attraction is largely the ability to use natural-gas infrastructure that already exists.For now, the distinction is simple: Bloom is capturing the economics today. FuelCell Energy is trying to prove it can do the same. Plug is largely playing a different fuel-cell market.
The Upstream Risk: Scandium
Bloom's manufacturing ramp raises a second question: what happens if the fuel cell hardware can be assembled quickly, but one of its inputs cannot scale with it?
The most discussed candidate is scandium.
Bloom's solid oxide cells use a zirconia ceramic electrolyte doped with scandium to improve conductivity. Scandium is produced in very small volumes globally, much of it as a by-product of other mining processes rather than from dedicated mines. China also plays an important role in the existing supply chain.
That became a live debate in July when a short seller argued that Bloom's longer-term production ambitions could eventually collide with the size of the scandium market.
Bloom disputes that conclusion. The company says it has enough non-Chinese material for current demand and visibility to support production well beyond today's manufacturing plans.
For now, the important point is what we do not see. There is no evidence that scandium is slowing Bloom's output today, and the risk is specific to Bloom's solid-oxide chemistry rather than fuel cells as a whole.
So scandium should be treated as a monitored upstream risk, not the explanation for Fuel Cell Power's current score.
If it ever begins limiting how quickly Bloom can expand, the nature of the constraint changes. Demand-led tightness would have turned into a physical materials bottleneck.
What Becomes Binding Next?
We ran the next question through Tessara's research workspace: if deployments scale into multiple gigawatts, what becomes binding next?
The evidence points to three gates.
Research Tab on Tessara. Highlights three possibilities if deployments scale.
Manufacturing capacity: Bloom’s ability to scale output may become the first constraint.
Gas infrastructure: Large campuses could face delays securing pipeline connections, firm supply and permits.
Electrical equipment: Fuel cells avoid the grid queue but still require transformers and on-site distribution systems. Tessara already reads Power Transformers as critically tight.
Fuel cells can bypass one power bottleneck without bypassing the rest of the power stack.
What Would Change the Read?
From here, only a few signals really matter.
Fuel Cell Power tightens if:
Bloom adds capacity and delivery times still stretch.
More gigawatt-scale frameworks convert into firm deployments.
FuelCell Energy begins converting its pipeline, broadening realised demand beyond Bloom.
Fuel Cell Power eases if:
Manufacturing capacity catches up without longer delivery times.
Large project pipelines fail to convert into orders.
Grid or turbine availability improves enough to reduce the premium customers will pay for speed.
Fuel cells became valuable because the rest of the power stack got slow. Demand is real, but manufacturing has so far responded quickly enough to prevent outright scarcity.
The next question is whether that balance survives another wave of gigawatt-scale deployments or whether the bottleneck moves into the materials, gas connections and transformers underneath them.
See where the AI buildout goes next
This issue, we identified on-site fuel-cell power as an increasingly valuable escape route from grid delays, Bloom Energy as the supplier already converting that demand into revenue, and the inputs and infrastructure that could constrain the next stage of deployment.
In Tessara, you can track Fuel Cell Power, Bloom’s upstream supply chain and the public companies exposed as the constraint changes.
See you next week,
Teng & Arvind
This article is for informational and research purposes only. It is not financial advice, investment advice, or a recommendation to buy or sell any security. Tessara Research does not publish price targets. The views expressed here reflect our analysis at the time of publication and may change as new evidence arrives. Readers should do their own research and consult a qualified financial adviser before making investment decisions.