CoreWeave is securing additional capacity at data centers owned by Core Scientific.
Cryptomine firm Core Scientific this week announced that GPU cloud provider CoreWeave is leasing a further 70MW from the company in a new 12-year deal.
Core Scientific will modify a total of 100MW of its owned infrastructure to deliver approximately 70MW to host CoreWeave’s Nvidia GPUs for HPC operations.
Site modifications are expected to begin in the second half of 2024, with operational status anticipated in the second half of 2025.
The announcement comes after CoreWeave exercised its first option to contract additional infrastructure under the terms of the 200MW, 12-year hosting contract announced earlier this month. This deal adds 70MW to the originally contracted 200MW.
“We are excited to build on our momentum and expand the scope of our HPC hosting business with significant additional infrastructure,” said Adam Sullivan, Core Scientific CEO. “The world is changing, and many data centers built in the last 20 years are not suitable to support future computing requirements. Our application-specific data centers will serve clients such as CoreWeave as they deploy next-generation chips with higher densities at scale.”
Core Scientific said the deal will add an additional $1.225 billion in projected cumulative revenue over the 12-year contract timeline, in addition to the $3.5 billion previously predicted from the original deal. The new agreement with CoreWeave also provides opportunities for two renewal terms of five years each.
In addition to the 270MW already contracted and set to be delivered by H2 2025, CoreWeave retains options for a further 230MW at other Core Scientific sites.
Both CoreWeave and Core Scientific were founded in 2017 as crypto firms.
CoreWeave – which counts Nvidia as an investor – pivoted to a GPU cloud offering, providing access to GPUs for AI applications. Core Scientific, meanwhile has continued to focus on hosting cryptomining hardware for itself and others, and is increasingly starting to host AI-focused hardware.
The duo have worked together for several years. Core Scientific said it hosted “thousands” of CoreWeave’s GPUs in its data centers from 2019 to 2022, and in March the companies announced CoreWeave was leasing 16MW of capacity from Core Scientific's Austin data center.
Under the 200MW deal announced earlier this month, Core Scientific said it will modify multiple existing, owned sites to host CoreWeave’s Nvidia GPUs. The crypto-company intends to redeploy Bitcoin mining capacity from designated HPC sites to its other dedicated mining sites.
CoreScientific said that with 1.2GW of contracted power, the company is able to deliver nearly 500MW of HPC capacity. The company operates cryptomining data center campuses in Texas, North Dakota, Kentucky, Georgia, and North Carolina.
CoreWeave, meanwhile, has raised billions of dollars in equity and billions more in debt financing as it looks to become a major player in the AI cloud space. It has been on a major leasing spree in the last 18 months and previously said that it expects to operate 14 data centers by the end of 2023 and 28 by the end of 2024.
The company currently lists three data center regions on its website; US East in Weehawken, New Jersey; US West in Las Vegas, Nevada; and US Central in Chicago, Illinois, while its status page also lists a region in Reno. It has signed leasing deals with multiple providers across the US and is expanding into Europe.
Core Scientific filed for bankruptcy in late 2022 and emerged from proceedings earlier this year. This month it rejected an offer from CoreWeave to buy the company.