Anthropic Signs $35 Billion Lambda Cloud Deal as AI Compute Financing Enters a New Phase

Anthropic has signed a cloud-computing agreement worth $35 billion with Lambda, a Nvidia-backed provider, to expand the infrastructure available for training and running its AI systems. The deal reflects the scale of computing required by frontier-model companies as demand for products such as Claude grows.

The agreement was described by people familiar with the matter and reported by Bloomberg, Reuters and The Wall Street Journal. Lambda markets dedicated, single-tenant infrastructure built around Nvidia GPUs, including H100, H200, B200, B300 and GB300 systems. The arrangement gives Anthropic a large external capacity channel while preserving access to specialized hardware beyond the company’s own facilities.

The project is tied to a data center in Nueces County, Texas, with planned capacity of about 350 megawatts, Reuters reported through Marketscreener. Hut 8, a company that shifted from cryptocurrency mining toward AI data centers, is developing the facility.

A 350-megawatt site ranks among the largest infrastructure commitments in the AI sector, linking model development directly to power procurement, cooling systems and industrial construction. Lambda’s cloud model centers on high-density power and liquid cooling, while Hut 8 supplies data-center development expertise. The physical buildout therefore matters as much as the contract value: Anthropic’s expansion depends on a facility capable of housing and powering large Nvidia GPU clusters.

Nvidia’s connection to the agreement extends beyond selling processors. The company backs Lambda, making the transaction part of a broader ecosystem in which Nvidia supplies hardware, supports infrastructure providers and benefits from the rising demand for compute.

The deal also illustrates how AI infrastructure is being financed through layered corporate and financial relationships. Coverage from Financialpost, Finance.yahoo and Capacityglobal described Google as supporting financing connected to Anthropic’s infrastructure commitments, while Apollo and Blackstone were linked to a separate $35 billion financing package for Anthropic’s chips. Those arrangements point to a market where cloud contracts, chip purchases, data-center leases and private credit are increasingly packaged together.

The Lambda agreement gives Anthropic access to additional Nvidia-based capacity at a time when leading AI companies are competing for scarce data-center power and advanced accelerators. Large commitments help providers build facilities with predictable demand, while customers secure infrastructure tailored to model training and inference.

For Anthropic, the arrangement broadens its infrastructure base and supports growth in both research and commercial services. For Lambda, a contract of this size provides an anchor customer for a major buildout. Nvidia gains another route for GPU deployment, while Google, Amazon and Microsoft face a stronger rival with the capital and capacity to scale. The financial structure also shifts risk across chipmakers, cloud providers, landlords and lenders rather than leaving one company to fund the entire expansion.

The $35 billion figure is part of a wider financing wave around Anthropic’s compute needs. Capacityglobal characterized an Apollo and Blackstone transaction as a record private-credit package backed by AI chips, while Quartz reported that Blackstone was pitching a second debt package that reached $36 billion in an early proposal.

Those figures describe related financing activity rather than a single identical transaction, but together they show how lenders are treating GPUs and contracted compute as financeable assets. Google’s backing adds another layer, since an early Anthropic investor is also connected to lease-payment support for data centers. The model turns long-term AI demand into an underwriting basis for hardware and facilities, a structure that will shape how quickly smaller providers can compete with the largest cloud companies.

The next test is execution: building the Texas facility, installing Nvidia systems and bringing the promised capacity online. The contract’s reported value, duration and delivery schedule were not fully disclosed in the available accounts, and the announcement was based on sources familiar with a transaction that had not received a detailed public release.

Anthropic’s deal will be watched for evidence that dedicated AI clouds can secure multibillion-dollar customers while financing power-intensive sites. Lambda’s ability to deliver the cluster will affect Anthropic’s model roadmap, Nvidia’s infrastructure demand and the economics of independent cloud providers. If the project proceeds as described, it will strengthen the case for treating compute capacity as a strategic asset rather than a routine cloud expense.