AMD EPYC and Instinct: Agreement to Provide 30 Megawatts of Computing Power to Rackspace

AMD and Rackspace Technology have announced the signing of a definitive agreement for the progressive deployment of an initial capacity of 30 MW dedicated to computing infrastructures based on AMD technologies within the provider's global data centers. The implementation will be distributed in multiple phases and will commence in the second half of 2026, with expected completion by 2028.

The agreement continues the memorandum of understanding communicated on May 7, 2026, and formalizes AMD's role as a strategic technology partner at the silicon level within Rackspace's AI governance stack. The goal is to build an enterprise cloud architecture capable of managing artificial intelligence workloads in regulated environments, with a centralized and end-to-end traceable operating model.

The overall capacity of 30 MW will be distributed across infrastructures based on AMD Instinct GPUs - including the MI355X and MI350P series, as well as future generations - and AMD EPYC CPUs, integrated into a unified Enterprise AI Cloud platform. The structure is designed to route workloads to the most suitable computing resources, balancing acceleration and general-purpose computing according to application needs.

According to both companies, the main focus is on regulated sectors, with particular attention to the healthcare sector, where the demand for scalable clinical inference and AI systems is growing. In these scenarios, the availability of controlled and verifiable infrastructures becomes an essential requirement, especially for applications that integrate into the core processes of organizations.

From an operational model perspective, Rackspace emphasizes that the goal is to provide a "governed" AI stack, where a single operator is responsible for both the infrastructure and operational outcomes, reducing the fragmentation typical of multi-vendor models. The approach aims to ensure greater consistency between performance, governance, and application outcomes.

AMD highlights the growing need from enterprise customers for infrastructures capable of flexibly combining AI acceleration and general-purpose resources. The integration with Rackspace's cloud model is intended to enable more efficient distribution of workloads, maintaining architectural openness and scalability.

The program also includes the coordinated use of commercial and go-to-market resources by both companies, with the aim of identifying and developing opportunities in regulated enterprise markets. The two companies will work together to capture demand for managed AI infrastructures, both for inference scenarios and for the deployment of AI agents in production environments.