2 hours ago
1
Rackspace Technology is going all-in on AI infrastructure, and it’s paying for the bet by cutting roughly 15% of its global workforce. The company formalized a definitive agreement with AMD to deploy an initial 30 megawatts of AMD-based AI compute across its global data center footprint, with rollout beginning late 2026 and extending through 2028.
RXT shares surged more than 25% in premarket trading following the announcements.
The deal and the cuts
The AMD partnership builds on a Memorandum of Understanding signed on May 7, 2026, and targets sectors where regulatory oversight is heaviest: healthcare, finance, and the public sector. The deployment will lean on AMD’s Instinct GPUs, EPYC CPUs, and ROCm software suite, essentially giving Rackspace a full-stack AI offering from silicon to outcomes.
The workforce reduction, approved by Rackspace’s board on June 10, 2026, is the other half of the equation. The company expects annualized savings between $75M and $85M from the headcount cuts, though it will eat one-time charges of $14M to $19M this year to make it happen.
Why regulated industries matter
CEO Gajen Kandiah, who joined in September 2025, has been steering the company toward what Rackspace calls its “Enterprise AI Cloud” strategy. The focus on regulated sectors is deliberate and strategically significant.
Most hyperscalers, think AWS, Azure, Google Cloud, offer AI compute at massive scale. But regulated industries often can’t just dump sensitive patient records or financial data into a public cloud environment and call it a day. They need governed infrastructure with clear accountability chains, audit trails, and compliance guarantees. By partnering with AMD and offering what it describes as “full-stack accountability from silicon to outcomes,” the company is pitching itself as the AI infrastructure provider for organizations that need more hand-holding than a hyperscaler typically provides.
What this means for investors
For AMD, the deal is a meaningful win in its ongoing battle against Nvidia for AI compute market share. Every high-profile enterprise deployment that runs on Instinct GPUs rather than Nvidia’s H100s or B200s chips away at the narrative that Nvidia is the only game in town for serious AI workloads.
One risk that investors should weigh carefully: the deployment timeline stretches through 2028. That’s a long runway in a market where AI hardware generations turn over roughly every 18 months. The AMD chips being deployed in late 2026 may face competitive pressure from next-generation silicon by the time the full 30 megawatts are online.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) · 