Bitcoin mining firm Core Scientific announced an updated agreement with AI hyperscaler CoreWeave, granting CoreWeave access to a total of 270 megawatts (MW) of infrastructure, according to a June 25 press release.

CoreWeave exercised its option to contract an additional 70 MW of infrastructure, adding to the initial 12-year agreement for 200 MW of high-performance computing (HPC) hosting contracts signed on June 3.

Core Scientific will modify 100 MW of infrastructure to provide the additional 70 MW needed to host CoreWeave’s NVIDIA GPUs.

The agreement also allows CoreWeave to opt for another 230 MW of infrastructure at other Core Scientific sites. If fully realized, this would make Core Scientific one of the largest data centers in the US.

Core Scientific plans to begin site modifications in the latter half of 2024 and aims to start operations in the latter half of 2025.

Projected revenue

Core Scientific anticipates $1.2 billion in cumulative revenue over the 12-year contract, plus an additional $3.5 billion from previous agreements with CoreWeave.

CoreWeave will fund the infrastructure modifications, and Core Scientific estimates that $105 million related to the additional contract will be credited against hosting payments, not exceeding 50% of monthly fees until fully repaid. There are also options for two five-year renewal terms.

A previous press release estimated a similar credit program at $300 million.

Broader plans

Core Scientific CEO Adam Sullivan stated that the firm is well-positioned to design and build application-specific data centers for clients like CoreWeave, focusing on high-density next-generation chip deployments.

Sullivan emphasized the importance of Core Scientific’s high-power infrastructure and experienced data center team, noting that the company delivered its Austin data center to CoreWeave 30 days ahead of schedule.

Currently, Core Scientific can provide 500 MW of its 1.2 gigawatts of contracted power to HPC hosting.