New research from Rystad Energy puts a number on something the industry has been watching build for a while: data center developers are increasingly choosing not to wait for the grid. Rystad projects the fuel cell market tied to data centers will grow roughly tenfold by 2030 — from about $2.8 billion in 2025 to around $30 billion — and this isn't a speculative forecast built on announcements. It's backed by a contracted order book of roughly 9 GW, including framework agreements with Oracle, AEP, Equinix, and Brookfield.
That's the headline. The more interesting story is why this is happening now, and what it says about where data center power strategy is heading.
The Grid Isn't Moving Fast Enough — And Everyone Knows It
This fits directly into the time-to-power problem the industry keeps running into. Rystad's research points to U.S. grid interconnection timelines that have roughly tripled since 2015, now stretching three to six years for large loads. For a hyperscaler trying to bring a gigawatt-scale campus online in a fraction of that time, a multi-year interconnection queue isn't a delay — it's a dealbreaker.
Rystad's modeling projects 10.4 GW of cumulative fuel cell demand from data centers between 2026 and 2030, and estimates that around 40% of projected 2030 U.S. data center capacity is likely to pursue dedicated on-site power generation rather than a conventional grid connection. That's not a niche hedge anymore — it's approaching a structural share of how the industry plans to power itself.
Why Fuel Cells Specifically
Fuel cells have a few characteristics that make them a good fit for this exact problem. They can be deployed much faster than a grid interconnection or a large gas plant. They run on natural gas today, which means a developer doesn't have to wait on a new fuel supply chain to get power flowing — and they can transition toward biogas, renewable natural gas, or hydrogen as those supply chains mature. They also produce lower on-site emissions than combustion alternatives, which matters for developers managing sustainability commitments alongside speed.
North America is expected to account for roughly 91% of installed global on-site power generation capacity, a combination Rystad attributes to grid delays, federal tax incentives, and an already-established domestic supply chain. In other words, this isn't just a technology story — it's a policy and infrastructure story landing in exactly the right place at the right time for U.S. developers.
Fuel cell manufacturers are scaling to meet it. Rystad projects aggregate operational and planned manufacturing output reaching 4 GW per year by 2030, up from 1.8 GW today.
The Concentration Risk Nobody's Talking About
Here's where this gets more interesting than a simple growth story. Solid oxide fuel cells (SOFC) have become the dominant technology for always-on data center power, accounting for roughly 53% of cumulative stationary deliveries to date. And within that technology, Bloom Energy holds virtually every primary-load SOFC contract currently in the visible order book.
That's a real concentration risk. If demand accelerates faster than one manufacturer's production capacity can scale, the whole timeline advantage that makes fuel cells attractive in the first place starts to erode.
It gets more specific than that. Bloom's SOFC technology depends on scandium, a critical metal used in its electrolyte chemistry. According to Rystad's analysis, if Bloom fully utilizes its planned 2 GW manufacturing expansion, its theoretical scandium requirement would approach the size of the entire current global scandium market — estimated at around 60 tonnes per year. China heavily controls that global supply chain. Competitors using alternative electrolyte chemistries don't carry the same exposure, which means a sustained scandium constraint could meaningfully reshape market share as the sector scales, not just squeeze costs at the margin.
Rystad still projects SOFC system costs falling 20-25% by 2030 — but notes that pace depends on manufacturers reducing costs across the full delivered system, not just the fuel cell stack itself.
What This Means for Developers
This is another data point for something we've said before on WattThe?!: there's no single solution to the AI power challenge. Fuel cells aren't replacing the grid, nuclear, natural gas, or renewables — they're becoming another tool developers use to close the gap between when they need power and when the grid can deliver it. The same logic that's pushing developers toward brownfield sites with existing transmission infrastructure is pushing them toward fuel cells: anything that shortens time-to-power gets serious consideration right now, even at a real cost premium.
But the scandium bottleneck is worth watching closely if you're evaluating fuel cells as part of a power strategy. A technology that looks like a clean way around grid delays today can become its own supply-chain constraint tomorrow if the underlying material inputs don't scale with it. That's not a reason to avoid fuel cells — it's a reason to ask hard questions about which manufacturer, which chemistry, and which supply chain a given deal actually depends on before signing a framework agreement.
The Bigger Picture
Access to power is becoming as strategically important as access to compute, and fuel cells are one more proof point. A $30 billion market growing from a $2.8 billion base in five years doesn't happen because a technology is nice to have — it happens because developers are structurally out of better options on the timeline that matters. Watching how this concentration risk plays out over the next 12-24 months will say a lot about whether fuel cells become a durable pillar of data center power strategy or a bottleneck of their own.
Written by Chris Kalowes, founder of WattThe?! — 15+ years in utility-scale battery energy storage (BESS), renewable energy, and AI infrastructure, across utilities, IPPs, EPCs, developers, and technology providers.
Source: Rystad Energy research and analysis, "Fuel cell investment by data centers set to grow tenfold, reaching $30 billion by 2030," June 2026.