Summary: On April 20, 2026, AI nuclear developer Fermi disclosed that CEO Toby Neugebauer and CFO Miles Everson left the same day, and its stock closed down 22%. The company announced a Fermi 2.0 restructuring, an Office of the CEO, and a permanent-CEO search. Project Matador remains a planned 17 GW AI campus in Amarillo, with Bloomberg-reported customer friction adding uncertainty.
A single Monday — April 20, 2026 — cost Fermi both of its top executives. The nuclear developer powering AI data centers, traded on NASDAQ and the London Stock Exchange under FRMI, disclosed without warning that Toby Neugebauer — co-founder and chief executive — and Miles Everson, the CFO, were leaving the same day. The same announcement rolled out a restructuring program named “Fermi 2.0.” The stock closed that day down 22 percent.
The company traces back to co-founders Rick Perry, the former US Energy Secretary, and Neugebauer; few public listings have chased the “AI power shortage” thesis harder: the pitch is a 17-gigawatt-class AI campus at Amarillo, Texas, assembled from batteries, nuclear, solar, grid power, and natural gas. And this week, the operational strain finally spilled into the open.
There is context behind the violent selloff. Power delivery has turned into the choke point for new AI capacity, so a developer that can credibly commit gigawatts to a timetable earns a valuation premium — and the moment its governance starts wobbling, that premium comes off fast.
One Day, Two Departures at the Top
The company’s own statement lays out the mechanics. Neugebauer gave up the CEO and chairman roles while holding onto his board seat. Everson stepped down as CFO and then won a directorship through designation rights held by the Melissa A. Neugebauer 2020 Trust. The chairman’s job went immediately to Marius Haas, until now lead independent director — a former President and Chief Commercial Officer at Dell who went on to co-found investment firm BayPine — and he created an “Office of the CEO.” Jacobo Ortiz Blanes, the former COO, and Anna Bofa, who has advised the board since the IPO and carries Google and Meta on her résumé, run it as co-presidents and answer directly to Haas. Finding a permanent CEO falls to Haas, Lee McIntire, and Cordel Robbin-Coker as the search committee, with Heidrick & Struggles running the process; the interim CFO seat remains in negotiation. Joining the board alongside them is Jeffrey S. Stein — who co-founded Breakpoint Advisory Partners.
This is a holding pattern more than a succession blueprint: the founder keeps his board chair, the interim setup is billed as furnishing “stable, experienced leadership” through the transition; no permanent CFO has signed. No matter what the release says, the double exit at the top is what professional investors at capital-hungry infrastructure firms keyed in on.
Project Matador: the 17 GW HyperGrid Plan in Amarillo
At the center sits Project Matador in Amarillo, Texas — the “HyperGrid”, a private campus planned at 17 GW that pairs a combined-cycle gas project ranking among the biggest in the country with a brand-new clean nuclear facility counting among the largest in the US, then tops up with batteries, solar, and power from the utility grid; hyperscaler customers are the target. The same reset makes Dallas the company’s headquarters and opens an office on the Matador site itself. Ties to the Texas Tech University System are also flagged as a Fermi 2.0 priority, with Chancellor Brandon Creighton’s endorsement that the project has “the potential to deliver generational impact” on record, and his note that the two sides are talking in good faith about extending some milestones written into the lease.
Where Bloomberg Found the Cracks
Drawing on Bloomberg, TechCrunch describes a Project Matador that has lurched through recent months, friction involving a key customer who stays unnamed among the setbacks. Place that beside the announcement’s own commitments — pledges to “memorialize binding commitments” covering the first client-tenants, and a courtship of strategic investors, client-tenants and sovereign funds included — and Fermi 2.0’s real assignment is plain: halting the trust bleed. The 22 percent one-day slide is simply the price tag the market hung, in real time, on “governance shock plus customer uncertainty.”
Fermi 2.0: Fresh Names, a New HQ, and Trust to Rebuild
The wider AI infrastructure market should read Fermi’s implosion as a reality check right on cue. On the energy side, the AI story’s investor-deck stage is over; the work has moved to engineering schedules and customer delivery, where any governance tremor, slipped milestone, or friction with customers now flows directly into the share price. Nuclear-for-AI has stopped being a story set in the 2030s — it is an operating project pipeline, reporting out every quarter, and that shrinks what risk can stay hidden, and for how long.
Teams running data centers and platforms get a concrete lesson: their capacity math rests on when new power truly lands, and treating the schedule risk of depending on one power vendor as a planning input mirrors designing around single-chip and single-cloud dependence in your architecture. Fermi’s one-day 22 percent is the market already billing for exactly that risk.