AMD has recently finalized a deal with Core Scientific, granting them access to over 500 megawatts of AI-ready data center capacity in the United States starting in 2027. This agreement has the potential to expand up to 2.5 gigawatts, allowing AMD customers to utilize the capacity for deploying AI systems using the company’s hardware and software.
The infrastructure provided by Core Scientific is targeted towards a variety of customers, including AI model developers, cloud providers, and enterprises that utilize AMD’s computing platform. Regulatory filings have outlined the allocation of the initial capacity, covering 529 MW across five Core Scientific facilities in Texas, Oklahoma, Alabama, and Georgia.
AMD has secured leases for 377 MW at facilities in Texas and Oklahoma, while another 152 MW has been leased by an unnamed neocloud at sites in Alabama and Georgia. These leases come with initial terms of 15 years and options for extensions, with AMD having the ability to reserve an additional 1,925 MW of capacity.
Core Scientific will be responsible for developing and operating the sites, collaborating with AMD on the physical infrastructure design and system deployment. The partnership focuses on the deployment of AMD Instinct GPUs, EPYC server processors, and ROCm software, although specific details on configurations have not been disclosed.
The agreement also includes AMD receiving warrants to purchase Core Scientific common stock, with the potential for significant revenue generation. Core Scientific’s extensive portfolio of AI-ready data centers is set to enhance access to infrastructure for deploying AMD AI solutions at scale, according to AMD’s senior vice-president.
In conclusion, the partnership between AMD and Core Scientific marks a significant step in expanding AMD’s infrastructure role, providing customers with access to cutting-edge AI data center capacity for deploying AMD solutions effectively.
