Anthropic locks $35B cloud deal

Anthropic has signed a cloud-computing deal worth about $35 billion with Lambda, the Nvidia-backed cloud provider, according to Reuters and other outlets citing people familiar with the matter. The agreement adds another large block of long-term compute for Claude and comes less than a week after Anthropic struck a separate roughly $45 billion capacity deal with Nscale. The back-to-back agreements show how Anthropic is locking in infrastructure as demand for Claude grows. On September 1, the company also announced Enterprise Frontier Safeguards, a new deployment option that lets eligible customers keep Claude activity data in cloud infrastructure they control rather than in Anthropic’s systems. ### Where does the Lambda capacity come from? (msn.com) Reuters reported that the Lambda deal covers a Texas data center project being developed in Nueces County by Hut 8, the crypto-mining company that has expanded into AI infrastructure. Yahoo Finance, citing the Wall Street Journal, said the arrangement links Anthropic’s compute demand with Lambda’s GPU financing and Hut 8’s 1-gigawatt Beacon Point campus in Texas. (anthropic.com) Forbes and other industry reports said Lambda is expected to provide capacity from a single Texas campus, while Nvidia’s role extends beyond supplying chips because it backs Lambda and is tied to the infrastructure stack around the project. Reuters described Lambda as Nvidia-backed; other reports said Nvidia is expected to hold the lease on the facility, though that detail has not been confirmed by Anthropic in a public statement. (channelnewsasia.com) ### Why does the $35 billion figure matter? The $35 billion figure matters because it follows a separate roughly $45 billion Anthropic agreement with Nscale disclosed on August 26. Bloomberg said that Nscale deal would give Anthropic computing power from Nscale’s flagship data center development in West Virginia, while CNBC reported the contract covered about 460 megawatts of capacity. (forbes.com) Taken together, the two agreements amount to roughly $80 billion in recently disclosed compute commitments. That total is based on the reported values of the Lambda and Nscale contracts and reflects how frontier AI companies are reserving capacity years in advance rather than relying only on short-term cloud purchases. (bloomberg.com) ### What did Anthropic launch alongside the infrastructure push? Anthropic said on September 1 that Enterprise Frontier Safeguards, or EFS, combines zero data retention with automated safeguards intended to detect misuse. The company said EFS works by storing data in cloud infrastructure controlled by the customer, not Anthropic. Help Net Security reported that the system allows Claude logs to remain in a customer’s own cloud under the customer’s keys while Anthropic’s automated systems continue to monitor for misuse. (msn.com) Anthropic said the product is being rolled out in phases and will be supported on Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google’s Agent Platform and Microsoft Foundry. (anthropic.com) ### Who is this aimed at? Anthropic said it built EFS with input from security, product, compliance and delivery teams at customers that need both privacy controls and frontier-model safeguards. SecurityWeek said the offering is aimed at enterprises that want zero data retention without giving up monitoring designed to catch misuse or autonomous misbehavior. (helpnetsecurity.com) That product rollout sits alongside Anthropic’s effort to secure more physical capacity for Claude. Anthropic said EFS will begin rolling out later this fall, while the reported Lambda and Nscale agreements point to a longer-term buildout of compute tied to named infrastructure partners including Lambda, Hut 8 and Nscale. (anthropic.com)