Rackspace, AMD to deploy 30 MW AI cloud for enterprises

Rackspace and AMD have signed an agreement for the phased deployment of 30 MW of AMD-based compute across Rackspace's global data centres, making AMD a strategic technology partner in Rackspace's AI offering.

The deployment is due to begin in late 2026 and continue through 2028. Rackspace plans to dedicate the initial footprint to AMD-based systems for enterprise AI workloads. The arrangement formalises an earlier memorandum and targets customers in regulated industries.

The infrastructure is intended to support workloads that require tighter governance and accountability, including projects in sectors such as healthcare. At full deployment, the 30 MW footprint is expected to provide dedicated compute capacity for organisations running AI models and inference tasks in sensitive operating environments.

The agreement covers AMD Instinct GPUs, including MI355X and MI350P, as well as future successor products, alongside AMD EPYC CPUs. Those chips will sit inside what Rackspace describes as an integrated Enterprise AI Cloud architecture, allowing workloads to be assigned to different types of compute depending on need.

The move adds a silicon supplier to Rackspace's broader effort to assemble a single AI stack for enterprises rather than rely on multiple separate vendors. Rackspace has already announced partnerships with Palantir and Uniphore, which it said would contribute data, inference and agent-based software layers to the wider offering.

That approach reflects a broader push among infrastructure and services providers to capture more of the enterprise AI budget by bundling hardware, software and operational management. For customers in regulated sectors, vendors are increasingly presenting governance, auditability and operational responsibility as key selling points as AI systems move beyond pilot projects and into production.

Regulated Demand

Healthcare providers were among the early organisations to show interest in access to accelerated compute for clinical AI and large-scale inference. Rackspace is positioning the new deployment as part of a service model for customers that want managed infrastructure rather than sourcing and integrating compute themselves.

In Rackspace's view, that demand is being driven by a shift from isolated AI experiments to agent-based workflows tied more closely to core business processes. This has become a common theme across the sector, as enterprises look for ways to run AI tools within existing systems while meeting internal controls and industry rules.

Gajen Kandiah, Chief Executive Officer of Rackspace, described the deal as a response to that market requirement.

"Enterprises in regulated industries need AI infrastructure that is governed from the ground up, with one operator accountable for business outcomes, not a collection of vendors each owning a piece," said Gajen Kandiah, Chief Executive Officer of Rackspace. "This collaboration combines the right compute with the right operating model and delivers something the market hasn't offered before: a governed AI stack with one accountable partner from silicon to outcomes."

Partner Strategy

AMD is seeking to strengthen its position in enterprise AI infrastructure as customers look beyond training large models and towards running mixed workloads in production. In AMD's view, enterprises increasingly need a combination of accelerated and general-purpose compute, rather than a single architecture for every use case.

Dan McNamara, Senior Vice President and General Manager of Compute and Enterprise AI at AMD, said the partnership was intended to meet that need.

"As enterprise AI evolves, customers need infrastructure that can deliver the right mix of accelerated and general-purpose compute for each workload," said Dan McNamara, Senior Vice President and General Manager of Compute and Enterprise AI at AMD. "By bringing together leadership AMD AI compute solutions and Rackspace's governed cloud operating model, we are helping regulated enterprises deploy high-performance AI infrastructure with the openness, scalability and accountability needed to run AI at enterprise scale."

Both companies said they would assign sales and marketing staff to pursue joint customer opportunities across regulated industries. The agreement is also intended to support four services previously outlined by the companies: Enterprise AI Cloud, Enterprise Inference Engine, Inference as a Service, and Bare Metal AMD Instinct.

Those services are meant to span the stack from underlying hardware to managed inference. Rackspace said the aim is to offer an alternative to bare metal procurement models by taking on operational responsibility for enterprise customers that want AI infrastructure delivered as a managed service.

The agreement also highlights how infrastructure providers are trying to define a clearer commercial role in enterprise AI beyond supplying raw compute. As adoption broadens, the contest is shifting towards who manages the systems, integrates the tools and carries accountability for running them in live business environments.

At full deployment, the footprint will amount to 30 MW of dedicated AMD compute across Rackspace's global data centre estate.