CoreWeave, supported by NVIDIA, has increased its AI hosting capacity with Core Scientific, adding 70 megawatts to their existing agreement.
CoreWeave, an AI cloud computing provider supported by NVIDIA, has recently increased its AI hosting capacity with Core Scientific through a new contract to add 70 megawatts to their existing deal. This expansion, announced on Tuesday, follows a previous 200-megawatt high-performance computing (HPC) hosting agreement the two firms entered last month, as reported by Core Scientific in a press release.
As per the new arrangement, Core Scientific will adapt 100 megawatts of its current infrastructure to accommodate CoreWeave's NVIDIA GPUs for HPC tasks. The modifications are scheduled to start in the second half of 2024, with the load anticipated to become operational in the latter half of 2025.
Core Scientific's CEO, Adam Sullivan, expressed enthusiasm for the expansion stating, "The world is changing, and many data centers built in the last 20 years are not suitable to support future computing requirements." He highlighted the company's strategic position to build application-specific data centers (ASDCs) tailored for the next generation of high-density chips.
This development is expected to contribute an additional $100 million in projected annual revenue for Core Scientific and estimated cumulative revenue of $1.225 billion over the 12-year contract. By the end of 2025, Core Scientific forecasts operating 270 MW of AI hosting capacity for CoreWeave, with a total revenue projection of $3.5 billion over the next 12 years.
CoreWeave will finance the capital expenditure required for the additional 70 MW expansion, which will be reimbursed through hosting fees once the project is live. Moreover, CoreWeave retains the option to further increase its infrastructure by another 230 MW at other Core Scientific sites.
