#AnthropicSigns35BCloudDeal

The $35B Anthropic deal is really a story about the physical economy behind AI.

Anthropic has reportedly signed a $35 billion cloud-computing agreement with Nvidia-backed Lambda for capacity at a data center being developed by Hut 8 in Nueces County, Texas. The project is expected to provide roughly 350 MW of capacity for Nvidia-powered computing aimed at supporting Anthropic’s Claude products and Claude Code.

The interesting part is the chain underneath the headline.

Anthropic needs compute → Lambda provides cloud capacity → Nvidia supplies the accelerators → Hut 8 develops the data-center infrastructure.

Reports also say Nvidia itself holds the lease on the facility, showing how deeply chip companies are becoming involved in the infrastructure layer of the AI economy. The exact financial arrangements between the parties have not been publicly disclosed, so it would be premature to treat the structure as guaranteed revenue for Nvidia or Hut 8.

For markets, the immediate beneficiaries are likely to be the AI infrastructure ecosystem rather than just AI software. Nvidia remains central, while data-center developers, power providers, networking, memory and cooling companies could all benefit if hyperscale compute demand continues expanding.

The timing is important too. Nvidia recently reported $89B in quarterly data-center revenue, up 117% year over year, and forecast another major revenue increase for its next quarter. That suggests the current AI infrastructure cycle is still being supported by exceptionally strong demand.

But there is another side.

A $35B commitment does not automatically mean $35B of immediate economic profit. AI infrastructure requires enormous capital, electricity and hardware, while accelerators can become economically obsolete as new generations arrive. The bigger risk is whether future AI revenue grows quickly enough to justify today's enormous infrastructure commitments.

Anthropic is also diversifying its compute sources. It recently agreed to spend $45B on Nscale capacity in West Virginia, while reports indicate the company is exploring longer-term alternatives to Nvidia hardware as well.

My takeaway: the AI race is moving beyond models and into land, electricity, chips and data centers. Watch $NVDA, $HUT, semiconductor demand, data-center power capacity and AI revenue growth.

The companies that control the compute supply chain may ultimately matter just as much as the companies building the models.

#GateEventContractTradeSharingChallenge
@GateSquare @Gate_Square