Rackspace Technology and Advanced Micro Devices signed a memorandum of understanding on May 7, 2026, to create a multiyear strategic partnership aimed at building an Enterprise AI Cloud for regulated enterprises and sovereign workloads. The plan is to combine AMD Instinct GPUs and EPYC CPUs with Rackspace’s governed operating model in a fully managed stack that the companies say is designed for security, governance and accountability.
The companies said the proposed platform would bring together four integrated capabilities: bare metal compute, developer-ready inference tooling, a fully operated inference runtime with defined service levels, and a governed Enterprise AI Cloud. Rackspace said it wants to own the stack from silicon to outcomes, while the cloud would be built on AMD Instinct GPUs, AMD EPYC CPUs and Rackspace’s managed environment.
Gajen Kandiah framed the deal as a response to a basic question from customers moving AI into production: who they can trust to run it. He said governance in regulated environments cannot be bolted on later and has to be built in from the start, adding that Rackspace and AMD are building a new category of enterprise AI infrastructure that the market has been asking for. Dan McNamara said enterprise AI is moving from experimentation to production and needs a compute foundation engineered for performance and efficiency at scale.
The announcement lands as many enterprises still rent GPU capacity by the hour and then shoulder the work of integration, security and accountability on their own. Rackspace is trying to wrap those pieces into one curated enterprise AI stack, and the AMD collaboration is intended to help it finish that effort. The centerpiece is the Enterprise Inference Engine, which would retain domain knowledge, session history and enterprise-specific data context across queries, with Rackspace owning the service-level commitment and taking responsibility for availability, scaling and performance.
That is the real shift in the deal. It is not simply more AI hardware on a rack; it is a bid to make one company answerable for the full path from chip to outcome in places where control matters most. If Rackspace and AMD can deliver that model, it could give regulated customers a cleaner way to deploy AI without stitching together their own stack piece by piece.
