GPU cloud company CoreWeave has exercised its final option to contract for additional infrastructure from Bitcoin and high-performance computing firm Core Scientific.
In June, Core Scientific announced a deal to host CoreWeave GPUs at its data centers, with this 120MW contract now bringing the total capacity to 500MW.
On the same day that it announced that first deal, Core Scientific turned down an acquisition offer from the AI cloud company, which it said “significantly undervalues” the business.
Core Scientific will modify the infrastructure at one of its sites to deliver the 120 incremental MW of critical IT load to host CoreWeave’s Nvidia GPUs.
The modifications are expected to begin in the second half of 2025, with the data center to be operational in the second half of 2026.
The 12-year HPC hosting contract is expected to add around $2 billion in additional cumulative revenue, along with $6.7 billion in cumulative revenue from previous CoreWeave contracts.
“In May, we communicated our plans to contract approximately 500MW of critical IT load to host high-performance computing. With today’s announcement, we have delivered on that commitment,” said Adam Sullivan, Core Scientific’s CEO.
“There is significant demand for infrastructure to support next-generation compute workloads. Our contracts with CoreWeave represent the foundation for Core Scientific’s evolution into a leading data center business ideally positioned to meet that demand.
“We are now working to expand power allocations at some of our existing data centers to increase our capacity for HPC hosting while also continuing to evaluate new sites in our pipeline to expand our business opportunity."
Core Scientific expects to have 500MW of infrastructure ready for CoreWeave by the second half of 2026. Another 500MW will continue to be used to mine Bitcoin.