Nscale, a company established just last year, has already secured major partnerships with names like Microsoft and is set to provide Anthropic with computing power built on Nvidia’s Vera Rubin system—the latest cutting-edge chip architecture. This system brings together six distinct chips operating in unison, widely regarded as the forefront of current chip technology. According to a source, this compute capacity is slated to begin supporting Anthropic's services toward the end of 2027. The arrangement, which was initially reported by Bloomberg, marks yet another move in Anthropic’s growing string of compute-focused collaborations.

Bloomberg notes that the agreement extends over a six-year period, with the computing resources coming from Nscale’s flagship data center located in West Virginia. Over the last eight months, Anthropic has been rapidly expanding its compute capabilities to sharpen its competitive edge, particularly against rivals like OpenAI. Just this month, Anthropic inked a $10 billion deal with Volta, an AI cloud startup founded only in January of this year, ensuring a six-year supply of cloud computing power sourced from a facility in Norway. In July, the company also entered into a $5 billion compute agreement with AMD. A few months prior, in May, Anthropic announced a substantial computing partnership with SpaceX—led by Elon Musk, whose ongoing friction with OpenAI CEO Sam Altman has positioned him as an unexpected supporter of Anthropic. That particular deal taps into computing capacity from two separate SpaceX data centers and reportedly delivers around $1.25 billion worth of compute resources each month.

In April, Anthropic further broadened its collaboration with Amazon, securing access to an additional 5 gigawatts of computing power. That same month, the company also deepened its ties with both Google and Broadcom, adding even more capacity to its arsenal. Anthropic is far from the only player aggressively chasing more AI processing strength. The wider industry is engaged in a relentless race to lock down as much compute capacity as possible, with other major entities—including Google, OpenAI, and Meta—pursuing similarly ambitious strategies.