πŸ”‹BESS
News Brief
iron-air batteries for data centers
energy storage technology
AI data centers
Form Energy

Can Iron-Air Batteries Power the AI Data Center Boom?

InfraSale Editorial
March 27, 2026
23 views
Energy Storage News

Could iron-air batteries be the key to powering the next generation of AI data centers? Discover the potential in our latest blog post!

The AI data center industry has a power problem, and it's getting worse faster than most people expected. GPU clusters hungry enough to train frontier models don't just need electricity β€” they need *guaranteed* electricity, available around the clock, independent of what the grid happens to be doing. That's a fundamentally different ask than what most energy infrastructure was built to deliver. This is why a 12GWh supply agreement signed at CERAWeek 2026 between Form Energy and data center developer Crusoe deserves more attention than another routine offtake announcement.

Iron-air batteries storing enough energy to power a small city for days at a stretch, manufactured in West Virginia, destined for AI campuses that may never touch the utility grid at all. That's not an incremental development; it's a glimpse at what energy infrastructure for the next decade of compute might actually look like.

What Iron-Air Batteries Actually Are

The chemistry is elegant in its simplicity, which is also why it took so long to commercialize. Iron-air batteries generate electricity through a controlled rusting process β€” iron oxidizes when exposed to air during discharge, then the reaction reverses during charging. The raw materials are about as abundant and cheap as they come: iron and air. No lithium, no cobalt, no manganese.

The tradeoff is duration versus efficiency. Lithium-ion systems can discharge and recharge with round-trip efficiencies (RTE) typically in the 85–95% range. Iron-air sits considerably lower. Form Energy has been transparent about this since emerging from stealth in 2021 β€” CEO Mateo Jaramillo has acknowledged it publicly rather than papering over it with marketing language. The honest framing is that iron-air isn't competing with lithium-ion; it's solving a problem lithium-ion was never designed for.

Where lithium-ion excels at short-duration applications β€” think 1-to-4-hour grid stabilization or peak shaving β€” iron-air is engineered for what the industry calls "multi-day" storage. Form Energy claims discharge durations exceeding 100 hours at full rated power. That's not a rounding error compared to lithium; it's a categorically different product.

The company's first commercial-scale validation came through a pilot project with Minnesota utility Great River Energy. Manufacturing is now underway at Form 1, a purpose-built factory in West Virginia, which matters for reasons beyond domestic supply chain optics β€” more on that shortly.

The Crusoe Deal: What 12GWh Actually Means

Twelve gigawatt-hours is a number that needs context. For reference, the entire U.S. grid-scale battery storage market installed roughly 10GWh in all of 2022. A single supply agreement committing to that volume β€” with secured pricing and delivery terms beginning in 2027 β€” signals that at least one major data center developer is treating multi-day storage as a serious infrastructure input, not a speculative bet.

Crusoe's positioning matters here. The company describes itself as an "energy-first" data center developer, explicitly prioritizing speed to power and low-carbon supply. In a market where hyperscale AI training campuses are competing for the same congested grid interconnection queues as solar farms and industrial customers, Crusoe's model of securing power alongside GPU capacity is a genuine competitive differentiator. The company that can promise a customer GPU compute with reliable, low-carbon power behind it is selling something fundamentally more valuable than raw compute alone.

The Crusoe agreement follows a much larger announcement: Form Energy's involvement in a 30GWh, 300MW storage system tied to a supply deal between Google and Minnesota utility Xcel Energy. That project β€” 100-hour duration dispatch, the largest energy storage project announced globally by watt-hour capacity β€” put Form Energy on the map as something beyond a promising startup. The Crusoe deal, at 12GWh, shows the pipeline is real and diversifying beyond utilities into private infrastructure developers.

Form Energy also recently announced its first international project, a 10MW/1,000MWh system with Irish renewables developer FuturEnergy Ireland, expected online in 2029. Three major announcements in a matter of weeks suggest the commercialization flywheel is starting to spin.

Why AI Data Centers Have a Multi-Day Storage Problem

The conventional assumption about data center power has always been: connect to the grid, buy firm capacity, maybe layer in some backup diesel generation. That model is breaking down under the weight of AI infrastructure demand.

Grid interconnection queues in the U.S. now stretch five to seven years in many regions. Utilities are under political and regulatory pressure not to let data centers β€” which can represent hundreds of megawatts of new load β€” crowd out residential and small commercial customers. The U.S. government has responded by actively encouraging "bring your own power" (BYOP) and "bring your own capacity" (BYOC) business models. Some data center projects are being planned as fully off-grid facilities specifically to sidestep interconnection constraints.

An off-grid AI campus running on renewables needs storage that can bridge not just overnight gaps, but multi-day lulls in wind and solar production. A single cloudy, calm week in winter isn't a hypothetical; it's a routine planning scenario for any serious energy engineer. Lithium-ion can handle hours. Iron-air, if Form Energy delivers on its specifications, can handle days.

The round-trip efficiency question is real, but it's also somewhat misapplied when critics raise it in this context. A data center operator sourcing power from dedicated wind or solar at low cost per MWh can absorb a lower RTE if it means eliminating the business risk of load curtailment or grid dependence entirely. The economics look different when the alternative is a seven-year interconnection wait or diesel backup at scale.

The Challenges Are Real β€” So Is the Opportunity

Skepticism about Form Energy is warranted, and intellectually honest observers shouldn't skip past it. The energy storage industry has seen multiple non-lithium battery companies accumulate impressive-looking gigawatt-hour supply pipelines that quietly evaporated before a single project reached commercial operation. Flow batteries, zinc-based systems, various thermal storage technologies β€” the graveyard of announced deals that went nowhere is real.

The Crusoe agreement is prospective. Delivery terms begin in 2027. Between now and then, Form Energy needs to scale manufacturing at Form 1, maintain cost targets, and execute on projects that are already in the pipeline. None of that is guaranteed.

The low RTE will continue to draw scrutiny, particularly as more efficient long-duration alternatives β€” including certain flow battery chemistries β€” improve. Form Energy's argument, that iron-air occupies a duration niche that no other technology serves at comparable economics, is coherent but untested at scale.

What the company has going for it: domestic manufacturing at a moment when U.S. policy strongly favors it, a chemistry based on materials that aren't supply-chain constrained, and a growing list of counterparties β€” utilities, hyperscalers, independent developers β€” willing to commit capital. The validation that matters most isn't the supply agreement announcement; it's the first 100-hour dispatch from a commercial project. That proof point, when it comes, will tell the industry everything.

For infrastructure investors and energy developers watching this space, the Crusoe-Form Energy deal is worth tracking less as a fait accompli and more as a leading indicator. The AI data center sector is going to need solutions that don't exist yet at scale, and multi-day iron-air storage is among the most credible candidates on the table. The projects that prove the model β€” or don't β€” will define which energy storage technologies get built at gigawatt-hour scale through the end of this decade.

The iron hasn't rusted yet. But the window for it to prove itself is opening fast.


Call to Action: Explore more about innovative energy solutions at InfraSale Marketplace.

[INTERNAL LINK: iron-air battery technology]

[INTERNAL LINK: AI data center energy solutions]

[INTERNAL LINK: energy storage market trends]

Related Topics:
energy storage technology
AI data centers
Form Energy

InfraSale Marketplace

Ready to act on this signal?

List a site or post a power requirement in under five minutes.