The AMD and Core Scientific logos
News

AMD & Core Scientific Announce $14B AI Data Center Deal

2 MINUTE READ|AI NewsAI News|Jul 28, 2026
Michelle Hawley avatar
By
SAVED
The multi-year agreement locks in more than 500 MW of US capacity for AMD's AI customers, with room to grow fivefold.

Key Takeaways

  • Core Scientific and AMD announced a strategic partnership covering more than 500 MW of US data center capacity beginning in 2027.
  • AMD holds an option to expand the partnership to as much as 2.5 GW, though no timetable or commitment for that scale has been disclosed.
  • The companies will jointly design physical infrastructure and deploy AMD Instinct GPUs, EPYC CPUs and the ROCm software platform.

Core Scientific and AMD announced a strategic AI infrastructure partnership on Tuesday covering more than 500 MW of US data center capacity beginning in 2027. AMD holds an option to expand to as much as 2.5 GW, though the companies did not say that additional capacity had been ordered or provide a timetable for reaching it.

Under the agreement, the companies will jointly design physical infrastructure and deploy AMD Instinct accelerators, EPYC server processors and the ROCm software stack for AI workloads. The capacity is meant to serve customers deploying AMD products, not AMD's own computing needs.

"AI deployments are accelerating rapidly, and bringing that compute online requires trusted infrastructure partners with the scale and power to support the next era of AI."

- Mathew Hein

SVP & Chief Strategy Officer, Corporate Development, AMD 

AMD-Core Scientific Partnership at a Glance

FeatureDetails
Initial capacity500+ MW across US facilities
Expansion optionUp to 2.5 GW (no timetable disclosed)
Technology deployedAMD Instinct GPUs, EPYC CPUs, ROCm software
Infrastructure designJoint collaboration on facility design
Target customersModel builders, cloud providers and enterprises using AMD products
Equity componentMarket-priced Core Scientific warrants issued to AMD
Start date2027

What the Deal Does and Doesn't Include

The companies did not identify any customers, cloud platforms or AI models expected to use the capacity, nor did they specify which versions of AMD's Instinct, EPYC or ROCm products will be deployed.

The agreement is structured as a strategic infrastructure relationship rather than a purchase order for a specified number of processors or accelerators. It lacks site-level allocations, construction costs, lease rates and specific revenue projections. Actual deployment will depend on construction progress and the commercial conditions governing future capacity.

AMD cited supply chain constraints, competition, export controls and software-hardware compatibility as factors that could affect the partnership's expected benefits.

Core Scientific's Pivot From Bitcoin Mining to AI

Core Scientific runs purpose-built data centers for high-density colocation and still earns bitcoin through its own mining operations and hosting services.

The company said the bulk of its revenue now comes from high-density colocation services, and it is repurposing the rest of its mining facilities as circumstances allow. Its portfolio spans 11 facilities across seven states.

Core Scientific had already announced a major Texas expansion before the AMD deal. In April, the company said its Pecos campus could reach about 1.5 GW of gross power and 1 GW of leasable capacity. The site had 300 MW of gross capacity supporting bitcoin mining and was being converted into an AI data center campus, with the first capacity expected to become available in early 2027 — matching the stated starting year for the AMD partnership.

"Our proven execution capabilities and ability to deliver high-density infrastructure at scale position us to support AMD's technology roadmap and grow our relationship meaningfully over time," Core Scientific CEO Adam Sullivan said.

Editor's Note: For more coverage on the latest AI infrastructure deals…

Main image: AMD, Core Scientific

About the Author

Michelle Hawley is Editorial Director at VKTR and host of The Inference. She covers the evolving AI landscape, including AI infrastructure, LLM development and enterprise AI strategy. With more than 10 years of experience, she has written for various publications, including The Press Enterprise and The Ladders, and taught courses on writing at Lycoming College.
Featured Research