Anthropic Signs $35 Billion Deal With Nvidia-Backed Lambda

Anthropic has signed another massive computing power deal. The maker of the Claude models has committed to paying $35 billion to Lambda, an Nvidia-backed cloud startup, for access to computing infrastructure over the next six years.

The news was first reported by Reuters and Bloomberg, citing people familiar with the details of the agreement. Lambda, a startup based in San Jose, California, will give Anthropic access to Nvidia chips installed in the new facility, and in return Anthropic will secure a guaranteed pool of computing power needed to train and run future versions of the Claude models.

Who's Building, Who's Paying

The structure of the deal has multiple layers. The data center in Nueces County is being built by Hut 8, a company that transformed from a bitcoin mining operation into an AI infrastructure operator. Nvidia leases the facility from Hut 8, and Lambda then installs the chips purchased by Nvidia and sells the computing power to Anthropic on a service basis.

Nvidia plays a double role in this arrangement, as an equity investor in both Anthropic and Lambda itself. This setup lets the chipmaker lock in demand for its own hardware while simultaneously funding the companies that buy and use that hardware.

Who Is Lambda

Lambda specializes in providing cloud computing optimized for training AI models, competing with bigger players like CoreWeave and Nebius. Backing from Nvidia has given it access to the latest chips at a time when global demand for GPU power exceeds supply, and AI labs are racing for every available batch of hardware.

The Race for Compute

The Lambda deal is another piece of Anthropic's much broader strategy. A week earlier, the company announced it would spend $45 billion to lease computing power at Nscale's data center campus in West Virginia. Counting from the start of 2026, the value of computing infrastructure deals signed by Anthropic has already reached at least $135 billion.

The scale of these commitments shows how costly it has become to stay at the front of the race for the best language models. Anthropic, like OpenAI and Google DeepMind, must secure access to computing power years in advance, since building new data centers and manufacturing chips takes years, while demand is growing faster than supply.

The company had earlier told investors ahead of its planned IPO that it sees a market opportunity worth as much as $30 trillion, which partly explains the scale of its current infrastructure spending. Without guaranteed computing power, it would be difficult to keep pace with developing successive generations of the Claude models against competition from OpenAI and Chinese labs.

What This Means for the Market

For Nvidia itself, the deal is further proof that its strategy of investing in customers who buy its own chips is paying off, though it also raises analysts' questions about cross-financial ties within the AI industry. For Hut 8, it confirms that its shift from cryptocurrency mining to AI infrastructure is delivering real multibillion-dollar contracts.

For readers in Poland, the key point is the scale of the broader trend. Energy, chip, and data center component prices are rising globally, partly driven by deals like this one, and the pressure on computing hardware supply chains is also being felt in Europe, where a parallel race is underway to build the continent's own AI gigafactories.