Rackspace Technology and AMD have moved past the memorandum stage and signed a definitive agreement to deploy 30 megawatts of AMD-based AI cloud compute across Rackspace data centers globally. The rollout runs in phases from late 2026 through 2028, and the customer base both companies are targeting is specific: regulated industries where AI cloud governance, data location, and operational accountability are not optional features but baseline requirements.
The AI cloud infrastructure centers on AMD Instinct GPUs, specifically MI355X and MI350X models, alongside AMD EPYC CPUs. All of it sits within Rackspace’s Enterprise AI Cloud architecture, which routes workloads across compute resources under a single cloud operating model rather than asking customers to manage separate infrastructure layers from multiple vendors independently.
That distinction is the actual pitch here. Rackspace CEO Gajen Kandiah described the target customer plainly: regulated enterprises need AI cloud infrastructure with one operator accountable for outcomes, not a collection of vendors each owning a separate piece of the cloud stack. For industries like healthcare, where early interest in clinical AI and large-scale inference is already visible, that kind of consolidated cloud accountability matters considerably when regulators start asking questions about data handling and system governance.
The agreement also formalizes four cloud service components that appeared in the earlier memorandum: Enterprise AI Cloud, Enterprise Inference Engine, Inference as a Service, and Bare Metal AMD Instinct. Together they cover the range from physical compute through fully managed cloud inference services, giving customers the option to engage at whichever cloud layer their operations require.
Both companies will assign dedicated sales and marketing staff to pursue regulated industry customers jointly, moving this beyond a typical hardware supply arrangement into a co-selling relationship with shared accountability for cloud outcomes.
For AMD, the deal deepens its enterprise AI cloud presence in a way that purely selling chips does not achieve. Tying its processors to a managed cloud operating environment gives AMD a stronger story for customers who care less about which GPU brand powers their cloud workloads and more about whether the entire AI cloud system meets their compliance requirements.
At full 30 megawatt deployment, the combined AI cloud footprint represents meaningful capacity for regulated enterprises moving workloads from pilot environments into core business systems.
