(Bloomberg) — Shares of Fermi Inc. plummeted as much as 46% on Friday after the power developer, co-founded by former Texas governor Rick Perry, announced that a tenant had terminated a $150 million agreement related to its proposed artificial intelligence campus in West Texas. The company revealed in a filing that the cancellation of the deal, originally signed on November 4, would have provided substantial funding for construction costs, although no funds had been drawn from this agreement.
The announcement has intensified fears of a potential bubble in the AI-powered energy sector, contributing to a decline in shares of several companies that had pledged to supply power to data centers. Fermi, which has yet to generate revenue, was valued at $19 billion just two months ago but has since lost more than two-thirds of its market capitalization.
Since its inception earlier this year, Fermi had enjoyed a significant rise, driven by its ambitious plan to develop the world’s largest private grid for a data-center campus — a project expected to consume more than twice the energy of New York City. “This was kind of the darling that came out in the hype of the AI power market, and right after the IPO, there were a couple of cracks that started to appear: Are we actually overbuilding these things and how are they being financed? The hype needs to shift into actual tangible results,” said Timm Schneider, an energy analyst and founder of The Schneider Capital Group.
Investor uncertainty is mounting over whether developers can secure long-term demand for massive new power projects aimed at supporting AI data centers. The focus is increasingly on which projects can convert initial interest into binding contracts. As a result, stocks linked to the AI boom experienced widespread declines on Friday, following a series of disappointing reports. BroadCom Inc., a major player in AI computing, failed to meet sales expectations, while Oracle postponed the completion of certain data centers it is developing for OpenAI. Shares of leading independent power producers, power-equipment manufacturers, and data center infrastructure developers also saw a downturn.
Despite the setback, Fermi maintains control of the Texas Tech land lease for its campus, known as Project Matador, as a previous letter of intent with the potential tenant remains in effect, according to Derek Soderberg, an analyst at Cantor Fitzgerald. Fermi and the former tenant are currently negotiating a lease agreement based on a letter of intent signed in September, which included an exclusivity provision that expired on December 9. The company has initiated discussions with several other prospective electricity customers and aims to commence power delivery in 2026.
The Texas developer indicated that the initial tenant attempted to impose last-minute pricing changes that were considered unacceptable. Negotiations are ongoing with two other prospective tenants. Soderberg noted that Fermi plans to deliver approximately 1.1 gigawatts of new power by the end of next year, equating to the output of a large nuclear reactor.
In the long term, Fermi envisions constructing four large nuclear reactors and has likened its efforts to the urgency of the Manhattan Project during World War II, which aimed to develop atomic bombs. The company faces escalating competition from established power developers and private equity firms, all vying for the same technology giants as clients.
As concerns about overbuilding and financing mount, Fermi’s future hinges on its ability to secure binding contracts and navigate the competitive landscape of energy development for AI infrastructure.
See also
Broadcom, Oracle Shares Drop Amid Intensifying AI Competition and High Investor Expectations
Intellect’s Arun Jain Unveils AI-Driven Banking Revolution at 2025 Global Bank Awards
AI Transforms Personal Finance: Edward Jones Highlights Opportunities and Limitations
Italian Firms Using AI Doubles to 16.4% in a Year, Yet 83.6% Remain Unengaged
Oversight Launches AI Platform to Monitor 100% of Enterprise Spend for Financial Risk Intelligence
