AMD and Rackspace sign 30 MW AI compute deal for regulated sectors
AMD and Rackspace Technology have signed a definitive agreement covering the phased deployment of 30 megawatts of AMD-based AI compute across Rackspace's global data-centre footprint. The rollout begins in late 2026 and runs through 2028, converting an earlier memorandum of understanding announced on 7 May 2026 into a binding commercial framework.
The compute layer will comprise AMD Instinct GPUs, including the MI355X and MI350P, alongside AMD EPYC CPUs, slotted into what Rackspace describes as an integrated Enterprise AI Cloud architecture. Rackspace says the setup allows workload-level routing across accelerated and general-purpose compute, with the operator taking accountability for end-to-end performance. Healthcare providers with early interest in clinical AI and inference at scale are cited as the initial target segment, though no named customers or committed contract values were disclosed.
The deal
The agreement operationalises four capabilities announced alongside the original MOU: Enterprise AI Cloud, Enterprise Inference Engine, Inference as a Service, and Bare Metal AMD Instinct. Both companies will assign dedicated sales and marketing personnel to jointly pursue customer opportunities in regulated industries. Specific pricing and financial parameters were described in the release's forward-looking-statements section as still subject to further agreement, and any third-party financing required remains conditional on acceptable terms.
Gajen Kandiah, chief executive of Rackspace Technology, framed the deal in terms of operator accountability rather than hardware configuration. "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," Kandiah said.
Market context
The partnership is a direct play in the fast-growing managed enterprise AI infrastructure category, where hyperscalers and a range of specialist managed-service providers are competing for budgets that are shifting from experimental AI pilots to production agentic workloads embedded in core business systems. Rackspace is positioning itself as an alternative to the bare-metal-plus-self-integration model offered by pure GPU cloud vendors such as CoreWeave and Lambda Labs, as well as hyperscaler IaaS offerings.
AMD, meanwhile, has been working to close the data-centre GPU market-share gap with Nvidia. Its Instinct MI300X series gained significant traction in 2024 and 2025 with hyperscale cloud operators; the Rackspace deal extends the company's reach into the managed enterprise and regulated-industry channel. For Rackspace, anchoring its AI stack to AMD compute is a strategic differentiator given that most managed cloud AI offerings default to Nvidia hardware.
Regulated sectors, including healthcare, financial services and government, represent a distinct buyer cohort with requirements around data sovereignty, audit trails and uptime guarantees that commodity cloud deployments often struggle to meet. Both companies have flagged digital sovereignty and governance as the primary commercial hook, which aligns with tightening EU and UK requirements around data residency for sensitive workloads under frameworks such as GDPR and the forthcoming EU AI Act's obligations on high-risk AI deployments.
Outlook
A 30 MW footprint, deployed across multiple geographies, is a credible entry-scale commitment for the managed enterprise AI market, though the phased structure and the outstanding financing conditions mean the full capacity realisation carries execution risk. Investors will be watching for named enterprise customer wins, clarification of the financing arrangements supporting the deployment, and any extension of the agreement beyond the initial 2028 horizon. Rackspace scheduled an investor call on 16 June to address questions from the market.