Anthropic signs $45B compute deal with Nscale for Nvidia Rubin chips
Anthropic has agreed to a $45 billion, six-year deal with Nscale to rent AI computing power from West Virginia using Nvidia Vera Rubin chips by late 2027.
Stock photo for illustration only, not from the actual event
- Anthropic has signed a $45 billion AI compute rental deal with British firm Nscale.
- The six-year agreement utilizes Nvidia's advanced Vera Rubin chip system from West Virginia.
- Compute capacity from the flagship data center is expected to power services by late 2027.
Anthropic is pushing forward with its aggressive expansion of computing infrastructure, reportedly signing an agreement to rent approximately $45 billion in AI compute from Nscale, a British artificial intelligence infrastructure company founded in 2024. The deal was confirmed to TechCrunch by a source familiar with the arrangement.
Under the six-year partnership, Nscale will supply Anthropic with computing power driven by Nvidia's Vera Rubin chips, a newly introduced state-of-the-art system that integrates six distinct chips to operate in unison. According to the source, this capacity sourced from Nscale's flagship data center in West Virginia is projected to begin powering the AI lab's services in late 2027.
Stock photo for illustration only, not from the actual event
This massive agreement represents just the latest move in Anthropic's relentless acquisition of computing partnerships over the past eight months, aimed at strengthening its competitive stance against rivals like OpenAI. The company previously secured a $10 billion deal with cloud startup Volta in August, a $5 billion partnership with AMD in July, alongside major infrastructure agreements with SpaceX, Amazon, and Google earlier in the year.
The intense race among leading artificial intelligence labs to secure massive computing power highlights how physical infrastructure and specialized chips have become the primary battleground in the generative AI industry. Partnerships spanning diverse infrastructure providers demonstrate the extreme capital expenditure required to scale next-generation foundation models.
Source: TechCrunch
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment