NVIDIA Holds Lease on Texas AI Facility Backing Anthropic's Cloud Expansion

Deep News15:53

Sources familiar with the matter indicate that AI firm Anthropic has entered into a six-year cloud computing agreement valued at roughly $35 billion with NVIDIA-backed cloud provider Lambda, aimed at rapidly scaling up computing power for products such as Claude. The infrastructure involved in this deal is situated in Nueces County, Texas, and is being constructed by Hut 8, a data center developer that transitioned from its origins as a Bitcoin mining operation.

Several weeks prior, NVIDIA secured capacity with Hut 8 and holds the lease on the facility. Lambda, in turn, obtained access to the machine room space from NVIDIA, deploying NVIDIA chips and reselling the computing capacity to Anthropic. Specific fees paid by Lambda to NVIDIA for the machine room usage have not been disclosed. As of the time of writing, Anthropic, NVIDIA, Lambda, and Hut 8 have all declined to comment on the matter.

It is important to clarify that this transaction does not represent a direct cloud contract from NVIDIA to Anthropic. Anthropic's direct contracting counterparty is Lambda. NVIDIA plays a threefold role in this deal: as an investor in Lambda, a chip supplier, and the holder of the machine room lease. For AI model companies that have not yet obtained an investment-grade rating, this structure ties machine room credit and chip supply to NVIDIA, mitigating contracting risks. For Lambda, it enables the signing of substantial contracts without the need to purchase an entire data center upfront.

One developer, two chip pathways

In July of this year, Hut 8 announced that an unnamed "high investment-grade customer" secured a 15-year lease on a roughly 700-megawatt campus in Texas, with a base lease contract value of approximately $19.6 billion. The facility was designed to support NVIDIA chips. On August 31, sources indicated that the campus in the same county is linked to NVIDIA's lease, though Hut 8 did not update the tenant name to NVIDIA in its announcement.

Meanwhile, sources note that Hut 8 is also developing another data center for Anthropic that will house Google's tensor processing units, with Google providing financial guarantees for those projects to assist Hut 8 in raising debt. This means that across different campuses under the same developer, Anthropic is utilizing NVIDIA chips via Lambda on one front, while pursuing the Google chip pathway on the other. The contract values and power-on timelines for the two routes should not be conflated.

Sequential capacity locks and NVIDIA's credit backing

Anthropic encountered tight computing supply earlier this year and has since signed a series of large-scale cloud contracts. Earlier this month, the company also reached a roughly $45 billion agreement with Nscale, another emerging NVIDIA-supported cloud provider, to lease NVIDIA capacity at a campus in West Virginia.

The $35 billion deal represents Anthropic's cloud procurement payment to Lambda and is distinct from the $19.6 billion 15-year lease on Hut 8's books; the two cannot be directly combined. NVIDIA announced in July that it would extend credit support to emerging cloud providers like Lambda in exchange for a share of cloud revenue. Some transactions under this program have recently been paused, and it remains unconfirmed whether Lambda is required to proportionally share revenue generated from the Texas campus with NVIDIA.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap
Leave a comment