Core Scientific, Inc. has extended its digital infrastructure agreement with CoreWeave, assigning additional resources for high-performance computing (HPC) operations within a 200-megawatt, 12-year contract that commenced on June 3, 2024. As per the recent update, Core Scientific has adapted 100 megawatts of its infrastructure to accommodate approximately 70 megawatts specifically for hosting CoreWeave's NVIDIA GPUs dedicated to HPC tasks. The renovation of the designated site is planned to start in the latter half of 2024, with operational functions expected to kick off in the latter half of 2025.
This newly established 12-year HPC hosting contract signifies a substantial increment in Core Scientific's exposure to extended, revenue-generating commitments denominated in dollars. The extension is forecasted to boost Core Scientific's anticipated revenue, anchored in multi-year, dollar-based agreements, by an estimated $1.225 billion over the consecutive 12-year period, supplementing the existing $3.5 billion derived from ongoing CoreWeave contracts.
Adam Sullivan, the Chief Executive Officer of Core Scientific, expressed enthusiasm about leveraging this fresh momentum to broaden the horizons of the HPC hosting division by incorporating significant supplementary infrastructure resources. Sullivan highlighted Core Scientific's strategic positioning to develop and establish application-specific data centers (ASDCs) characterized by heightened rack energy densities spread across extensive areas. These ASDCs are envisioned to provide support to clients like CoreWeave in implementing cutting-edge chips on a large scale, capitalizing on their adaptable, high-powered infrastructure and proficient workforce to ensure streamlined operations.
Furthermore, Sullivan emphasized core Scientific's intention to expedite the growth of its hosting arm by advancing project timelines, exemplified by the prompt execution of the Austin data center development for CoreWeave. The objective is to reinforce AI computational capabilities and seize the opportunities arising in bitcoin mining within the backdrop of ongoing industry transitions.
Under the terms of the agreement, CoreWeave will fund all fundamental upgrades required to modernize Core Scientific's densely packed HPC infrastructure, amounting to approximately $105 million. The expenses will be offset from the hosting charges, with a ceiling of 50% of the monthly payments until complete reimbursement. Additionally, the deal incorporates provisions for two successive renewal terms of five years each.
Core Scientific aims to expand its HPC infrastructure dedicated to CoreWeave to reach 270 megawatts by the end of 2025, building upon its existing capacity of 200 megawatts. Furthermore, CoreWeave retains the option to incorporate an additional 230 megawatts at alternative Core Scientific locations, potentially solidifying Core Scientific's standing as one of the prominent data center operators in the United States.
In summary, Core Scientific's contracted power capacity of 1.2 gigawatts positions it to furnish close to 500 megawatts for HPC hosting, catering to a diverse array of computational workloads predicated on geographical proximity.
