Fermi America signed a credit support agreement with AMD supporting a customer lease for the first phase of Project Matador, a large AI and energy infrastructure development in Carson County, Texas. The initial phase represents 222 MW of gross power capacity. The companies also outlined a framework for potential expansion to as much as 650 MW.

Project Matador is planned as a behind-the-meter AI data center campus with a private electrical grid. Fermi says it has invested more than $1.5 billion in the project to date. Its long-term power strategy includes natural gas generation, planned nuclear generation, utility supply, solar and battery storage.

AMD’s credit support is intended to facilitate financing for the initial customer lease. The framework for possible expansion does not establish that the entire 650 MW has been contracted, financed or built. The company has described a longer-term 17 GW campus vision, conditional on customer commitments, capital and development milestones.

The agreement illustrates the increasingly close relationship between AI compute procurement, power infrastructure, customer commitments and financing arrangements.

Analysis

Fermi America is a relatively new entrant in the AI infrastructure market, founded in January 2025 by energy entrepreneur Toby Neugebauer and former Texas Governor Rick Perry. The company completed a $785 million IPO in October 2025, just nine months after its formation, and is pursuing one of the industry’s most ambitious integrated power and data center developments. Its flagship Project Matador in the Texas Panhandle envisions a private electrical grid supporting AI data centers, initially using natural gas generation with longer-term plans incorporating nuclear power. Fermi has outlined a potential 17 GW development, although that figure remains conditional on additional land, permits, financing and customer commitments. The strategy is to address one of AI infrastructure’s largest constraints: access to substantial amounts of reliable electricity without waiting years for conventional utility interconnections.

Fermi has achieved several measurable milestones, including securing land, beginning site construction, obtaining a Texas air permit for approximately 6 GW of natural gas generation, procuring gas turbines and advancing federal licensing for four proposed AP1000 nuclear reactors. Its August 2026 agreement with TensorWave established a 15-year lease for 222 MW of gross power capacity, representing approximately $6.5 billion in contracted revenue over the initial term, with phased delivery expected to begin in late 2027. AMD’s credit support strengthens the financing framework for this initial development, while the companies are exploring expansion toward 650 MW. However, the distinction between permitted generating capacity, equipment procurement, contracted data center capacity and operational infrastructure remains critical. The broader 17 GW vision is substantially larger than the currently contracted first phase. (⁠Fermi America)

The arrangement also illustrates the potential risks of circular financing in the AI infrastructure ecosystem. AMD is both an investor in TensorWave and the supplier of the GPUs underpinning its AI cloud business, while now providing credit support for TensorWave’s lease obligations to Fermi. Such arrangements can accelerate infrastructure deployment by improving financing terms and aligning suppliers, cloud operators and developers. However, they also create interconnected financial exposures: if customer demand, GPU utilization or cloud revenues fall short of projections, the consequences could extend across multiple participants. Credit support does not itself establish that financing has closed or that future demand is secured. For Fermi, the decisive tests will be completing project financing, delivering reliable power and operational data center capacity on schedule, and demonstrating that TensorWave’s long-term lease obligations are supported by sustainable end-customer demand rather than continued capital commitments from the same AI infrastructure ecosystem.

We’re tracking AI infrastructure developments. Follow ongoing coverage at https://convergedigest.com/category/ai-infra/