**AMD Secures Over 500 MW of AI-Ready Data Centre Capacity with Core Scientific**
Advanced Micro Devices (AMD) has entered into a significant agreement with Core Scientific to access more than 500 megawatts (MW) of AI-ready data centre capacity across the United States, with plans beginning in 2027. The collaboration is part of AMD’s broader push to strengthen its infrastructure offerings and support the growing demand for scalable, high-performance computing driven by artificial intelligence (AI).
Under the initial agreement, AMD will have access to 529 MW of critical IT capacity spread across five Core Scientific facilities located in Texas, Oklahoma, Alabama, and Georgia. The arrangement includes direct leases and partnerships with other cloud operators, giving AMD flexibility in deploying its hardware and software solutions. The capacity is expected to support a wide range of customers, including AI model developers, cloud service providers, and large enterprises leveraging AMD’s computing platform.
**Capacity Secured Across Five US Sites**
The deals involve long-term leases, with initial terms of 15 years and options for three additional five-year extensions. AMD has reserved substantial room for expansion, with the potential to scale the agreement up to 2.5 gigawatts (GW) by 2028. The first phase of the initiative is anticipated to become operational starting in 2027.
Core Scientific is responsible for developing and operating the data centre sites, working closely with AMD on infrastructure design and system deployment. While the partnership focuses on deploying AMD’s Instinct GPUs, EPYC server processors, and ROCm software, specific technical configurations—such as processor models, networking systems, and cooling technologies—have not yet been disclosed.
The facilities will emphasize high-density AI deployments, with the 529 MW figure referring to critical IT capacity rather than total power consumption. Core Scientific’s regulatory filings also highlight that AMD will retain certain protections regarding equipment used at facilities leased by third-party neocloud partners.
**AMD Broadens Its Infrastructure Role**
Beyond securing data centre capacity, the agreement reinforces AMD’s expanding role in the infrastructure stack. AMD is integrating its hardware and software capabilities into a unified architecture, including its Helios reference design. Helios combines Instinct GPUs, EPYC CPUs, Pensando networking, ROCm software, liquid cooling, and rack-level power distribution into a cohesive system. Although it is unclear whether Helios will be deployed at the Core Scientific sites, the architecture reflects AMD’s strategy to deliver fully optimized solutions for AI workloads.
Core Scientific operates 11 facilities across multiple U.S. states and is actively repurposing former cryptocurrency mining sites into high-density colocation centres. During the second quarter of 2026, high-density colocation accounted for the majority of the company’s revenue, underscoring the growing importance of this segment.
Following the announcement, AMD reported that it has access to approximately 1.1 GW of leased customer power capacity, with a significant portion already billable. The company believes that the Core Scientific partnership will play a key role in supporting its product roadmap and accelerating AI adoption at scale.
“As Core Scientific’s extensive portfolio of AI-ready data centres expands access to the infrastructure our customers need to deploy AMD AI solutions at scale,” said Mathew Hein, AMD’s senior vice-president and chief strategy officer of corporate development.
—
### FAQ
**Q: What is the capacity agreement between AMD and Core Scientific?**
AMD has secured access to over 500 MW of AI-ready data centre capacity through Core Scientific, with potential expansion up to 2.5 GW by 2028.
**Q: Where are the data centre sites located?**
The facilities are located in Texas, Oklahoma, Alabama, and Georgia.
**Q: What technologies will be deployed under this agreement?**
The partnership focuses on AMD Instinct GPUs, EPYC server processors, and ROCm software, though specific models and configurations have not been disclosed.
**Q: What is Helios architecture?**
Helios is AMD’s reference architecture that integrates compute, networking, power distribution, and cooling into a unified rack-scale system designed for AI workloads.
**Q: Are AMD’s own facilities part of this agreement?**
No, the leased capacity is provided by Core Scientific. AMD is not acquiring ownership of the data centre sites.
**Q: When will the capacity become available?**
The first phase of the agreement is expected to be operational from 2027.
—
### Conclusion
AMD’s agreement with Core Scientific represents a major step in its infrastructure expansion strategy. By securing access to hundreds of megawatts of AI-ready data centre capacity across multiple U.S. locations, AMD is positioning itself to better serve enterprises and AI developers seeking scalable, high-performance computing solutions. Combined with its Helios architecture and growing data centre footprint, the partnership strengthens AMD’s role as a key competitor in the evolving AI hardware and cloud infrastructure market.
