CEO of Fermi America, co-founded by former Trump Energy Secretary Rick Perry, suddenly resigns; stock plummets nearly 30% after hours
Toby Neugebauer, CEO of Fermi America, has suddenly resigned, effective April 17, according to SEC filings. The company's stock plummeted nearly 30% after hours, with a total decline of 75% over six months. The company was co-founded by former Trump Energy Secretary Rick Perry and went public in October last year. It is developing a massive AI data center called the "Matador Project" in Texas, covering approximately 5,800 acres and planning for 17GW of power.
The project is stalled at two major hurdles: the lack of anchored tenants and unresolved cooling design. The March financial report acknowledged that supply chain bottlenecks exceeded expectations, and a Cleanview report stated that the first facilities may not be operational until May 2027. Co-founder Griffin Perry recently reduced his stake by 15%, and the exit of one tenant has triggered a collective lawsuit.
Source: Public information
ABAB AI Insight
CEO's sudden departure and stock crash expose the vulnerability of the AI data center "demand pre-sale" model. High valuations rely on future leases, but hyperscalers (like Amazon, rumored) require immediate power and custom cooling; if demand validation fails, the project faces a cash flow cliff.
The 17GW scale is equivalent to the power consumption of a medium-sized city, relying on a gas-nuclear mix, but global shortages in the supply chain (turbines, cooling towers, transformers) create a "chicken-and-egg" dilemma: no tenants, no investment; no power, no tenants. Although Texas has strong policy support, environmental assessments and grid access remain hard constraints.
The stake reduction and lawsuit reflect a collapse of internal confidence. The Perry family's stake reduction signals capital outflow, and the lawsuit amplifies uncertainty. This is not an isolated case but a typical example of the AI infrastructure bubble: political endorsement masks business risks, and after the IPO, execution capability is lacking.
In the long run, the massive park model faces differentiation: successful players monopolize power resources, while failures become "ghost towns." The real winners will be flexible, small-scale, tenant-pre-bound projects, proving that scale is not the only path to AI computing power expansion.