- Fermi America, the nuclear-powered AI data center startup co-founded by former Energy Secretary Rick Perry, went public in October 2025 with a Trump-branded campus near Amarillo, Texas.
- Six months later, the company has zero confirmed clients, its CEO and CFO have departed, shares tanked 22% in a day, and the co-founder is now suing the company for wrongful termination.
- The problem? No hyperscaler anchor tenant means no project—and no revenue.
Fermi America’s pitch was audacious: build the “Donald J. Trump Advanced Energy and Intelligence Campus” near Amarillo, Texas, a nuclear-powered data center campus that would deliver gigawatt-scale power to AI hyperscalers. Investors bought it—the company went public in October 2025, with Rick Perry, former U.S. Energy Secretary under Trump, as co-founder.
Six months later, the project is stalled, leadership has fled, and the stock is cratering.
The core problem is brutally simple: Fermi never signed a single client.
The Anchor Tenant Problem
Data center projects of this scale require an anchor tenant—typically a hyperscaler like Amazon, Google, Microsoft, or Meta—to commit to long-term capacity before construction begins. Without that commitment, financing dries up, timelines slip, and projects die.
Fermi has no publicly confirmed anchor tenant.
An independent report by Cleanview, shared exclusively with Axios, estimates that even if Fermi secured an anchor tenant this month and matched industry-standard construction timelines, its first buildings wouldn’t come online until May 2027—about a year later than initially projected.
CEO Toby Neugebauer acknowledged to Axios that the project “can’t move forward without a tenant,” though he also claimed the lack of clients “isn’t a problem” for the company. The market disagreed.
On April 20, 2026, Fermi announced that co-founder and CEO Toby Neugebauer and CFO Miles Everson had suddenly departed. Shares dropped 22% that day.
The company framed the departures as part of “Fermi 2.0,” a strategic reset that included moving corporate headquarters to Dallas. But the timing—coming just days after Axios reported on the lack of anchor tenant—suggested otherwise.
Ten days later, on May 1, 2026, Neugebauer sued Fermi and three board directors in Texas Business Court, alleging he was wrongfully terminated. His petition claims that under Texas law and Fermi’s organizational documents, only shareholders—not the company or its board—can remove a director. The company said Neugebauer was terminated for “conduct in violation” of his employment agreement.
The Trump Brand and AI Hype
Fermi’s Trump-branded campus was designed to cash in on the AI boom sweeping through the U.S. economy. When the company announced plans for the “Donald J. Trump Advanced Energy and Intelligence Campus” near Amarillo, investors clamored for a piece of the action.
The company’s public filings describe its mission: “strengthen national security, establish America’s energy dominance, and power AI innovation through its Project Matador campus.” It’s developing “next-generation private electric grids that deliver highly redundant power at gigawatt scale to support next-generation intelligence and AI compute.”
The rhetoric was compelling. The reality was not.
What Went Wrong
Fermi’s failure exposes a fundamental challenge in the AI infrastructure boom: building before you have customers is a recipe for disaster.
The company went public with a vision but no contracts. It promised gigawatt-scale power but couldn’t secure a single hyperscaler commitment. It touted national security and energy dominance but couldn’t convince a single tech giant to sign on the dotted line.
The contrast with successful AI infrastructure plays is stark. Applied Digital landed a $7.5 billion hyperscaler deal for its fifth AI campus. Hyperscalers are signing multi-billion dollar commitments with data center providers that have proven track records and existing infrastructure.
Fermi had neither.
Fermi’s struggles raise questions about the sustainability of the wider AI infrastructure boom. As Warren Buffett recently warned, AI debt could trigger “another 2008″—”I know a bubble when I see one.”
The company’s stock is publicly traded on NASDAQ and LSE under ticker FRMI. Its market cap has plummeted as investors realize that without anchor tenants, the Trump-branded campus is nothing more than a PowerPoint presentation.
The lesson is clear: in the AI infrastructure race, customers come first. Build it, and they will come? Not anymore.