Image to help understand the article
Nvidia is considering providing billions of dollars in credit guarantees to support construction of a U.S. data center campus that will supply computing capacity to artificial intelligence company Anthropic, The Information reported Oct. 29, citing multiple sources familiar with the discussions.
The proposed backing would cover the first building at British cloud infrastructure company Nscale's Monarch Compute Campus in West Virginia. The facility, designed to deliver 460 megawatts of computing capacity, is expected to house Nvidia's next-generation Vera Rubin GPUs. No agreement has been signed, and the terms remain subject to change.
A Different Kind of Financial Guarantee
Unlike previous arrangements in which Nvidia guaranteed customers' lease payments or GPU purchases, the proposal would protect the residual value of the data center building itself.
Under the structure being considered, Nvidia would repay creditors if Nscale defaulted on its construction financing. Nvidia could then take ownership of the facility and either sell it or lease it to another customer.
Industry estimates put the basic construction cost of a 1-gigawatt data center at approximately $12 billion. Based on that benchmark, the potential guarantee could exceed $5 billion, according to estimates cited in the report. Nscale hopes the backing would help it secure construction financing on more favorable terms.
Anthropic Contract Raises Stakes for Nscale
Nscale has already signed a $45 billion agreement with Anthropic to provide computing resources from its first data center over an extended period. The companies are also reportedly discussing similarly sized agreements for additional buildings.
However, Nscale has yet to secure the financing needed for construction and equipment at the first facility. According to an EY estimate commissioned by Nscale, developing the first 2 gigawatts of the Monarch campus would cost $69.2 billion.
Nscale reported revenue of $141 million and a net loss of $1 billion in the first half of 2026. Its contracted revenue backlog stands at $103 billion, representing payments expected from long-term service agreements. Anthropic accounts for more than 40% of that total. If Nscale fails to raise the necessary funding within the required timeframe, the agreement could be terminated.
Environmental and Infrastructure Challenges
The West Virginia project also faces regulatory and environmental hurdles. Nscale originally planned to use hydrogen power but switched to natural gas generation. The first 2 gigawatts of development are expected to include 864 gas-fired generators, but the company has not yet received final permits for air pollutant emissions.
Potential annual greenhouse gas emissions of up to 10 million metric tons and flooding damage in nearby communities during construction have also drawn controversy.
The project's risks extend beyond West Virginia. Microsoft recently withdrew from an Nscale data center project in Essex, England, citing delays in electricity supply. The area is not expected to have access to the necessary grid connection until sometime in the 2030s.
Nvidia's Expanding Financial Exposure to AI Infrastructure
Founded in 2024, Nscale is pursuing a U.S. stock market listing with a targeted valuation of $35 billion. The company was valued at $14.6 billion during a fundraising round in March.
Nvidia has invested in or provided credit support to so-called neocloud companies, including CoreWeave and Nscale. Its total credit support and investments provided or committed to these businesses exceed $170 billion, a figure that includes future guarantees and other commitments.
The strategy highlights Nvidia's financial involvement in the infrastructure supporting the AI industry. While major cloud providers such as Google, Microsoft and Amazon are expanding efforts to develop their own chips and reduce dependence on Nvidia GPUs, neocloud companies remain major buyers of Nvidia's processors. The proposed West Virginia guarantee would deepen that relationship by helping finance the facilities needed to deploy Nvidia hardware at scale.
Comments
Post a Comment