Rackspace Technology reported second-quarter revenue of $670 million, a rise of just 1% year-on-year, and used the results to unveil an expanded enterprise artificial intelligence infrastructure strategy anchored around managed compute, inference and sovereign-AI deployments.
Numbers in the quarter
The company said it continued to hold its full-year guidance and expects to deliver positive free cash flow for 2026 in a range of $50 million to $70 million. Public Cloud revenue declined, while Private Cloud grew.
| Metric | Q2 figure | Year-on-year change |
|---|---|---|
| Total revenue | $670m | +1% |
| Private Cloud | $263m | +5% |
| Public Cloud | $407m | -2% |
What Rackspace is planning for enterprise AI
The company set out a strategy to be an "accountable provider and operator of the full enterprise AI stack from core to cloud to edge," stressing governance, security, sovereignty, resilience and operational accountability for customers. It said a Managed Compute and Inference Platform will be backed by a partner ecosystem that includes AMD, Dell, Palantir and Uniphore.
"...an accountable provider and operator of the full enterprise AI stack from core to cloud to edge,"
A central plank of the plan is a definitive agreement with AMD to deploy an initial 30 megawatts of AMD-based compute capacity across Rackspace data centres in phased builds from late 2026 through 2028. The architecture will use AMD Instinct GPUs and EPYC CPUs, Rackspace said.
At full planned capacity the company projects the initiative could produce annual revenues in a band of $450 million to $600 million, with enterprise-AI EBITDA margins it expects to be above 50%. Those figures frame the commercial potential Rackspace sees in offering managed AI compute alongside higher-value services such as inference and sovereign-AI deployments.
Why it matters
The announcement positions Rackspace to capture work that requires tight governance and placement of workloads across on-premises, cloud and edge environments — areas where enterprises are increasingly cautious about latency, cost and data sovereignty.
- By pairing large-scale GPU capacity with managed services, Rackspace aims to move up the value chain from commodity public-cloud hosting to specialised AI operations.
- The AMD partnership and phased 30MW plan mark a concrete capital and technology commitment, not just product positioning.
- Improved Public Cloud margins suggest Rackspace is shifting its mix toward higher-margin AI, data and managed-services work despite a slight revenue fall in that division.
Private Cloud revenue rose to $263m, helped by contract timing, while Public Cloud fell to $407m. Rackspace said the Public Cloud decline was offset by margin improvement as the company refocused on more specialised workloads.
Context and competition
Cloud and infrastructure providers are racing to provide AI-optimised compute and end-to-end services for enterprise customers. Rackspace's approach emphasises an accountable, managed-service model intended to appeal to organisations that need strong governance, security and the ability to keep some workloads close to the enterprise or within national boundaries.
The scale Rackspace is targeting — tens of megawatts of GPU compute — follows broader industry moves to build out large, purpose-built clusters for generative AI and inference. The company is betting that enterprises will prefer a managed, hybrid option combining in-house and hosted resources rather than relying solely on hyperscalers.
While the financial upside Rackspace projects is significant, execution will require the company to build or retrofit data-centre capacity, integrate AMD hardware, and convert enterprise interest into multi-year contracts that justify the capital and operating expense.
For now, Rackspace's Q2 results show modest top-line growth, but the new AI infrastructure strategy supplies a clear playbook for where the company intends to find higher-margin growth in the years ahead.
