Vultr AMD partnership anchors its cloud AI infrastructure pitch
Vultr says AMD CPUs and GPUs, 33 regions and open stacks can undercut hyperscalers for enterprise AI workloads.
By Wei-Lin Zhao · AI Correspondent
· 3 min read
Vultr is pitching the Vultr AMD partnership as the hardware and economics base for its cloud AI infrastructure strategy, with AMD EPYC CPUs and Instinct GPUs supporting a public cloud footprint across 33 regions. Kevin Cochrane, Vultr Holdings LLC’s chief marketing officer, told SiliconANGLE’s theCUBE that the company is using AMD’s product roadmap, distributed data centers and open software stacks to compete for enterprise AI workloads.
The financial terms of Vultr’s AMD arrangement were not disclosed. Cochrane said Vultr’s AMD-based offering can deliver up to 33% better performance at an average of 82% lower cost than rival hyperscaler services. Those figures were presented as Vultr’s claim, and the company did not provide the underlying benchmark detail in the remarks reported from the AMD Advancing AI event.
Vultr’s argument is aimed at a shift in AI infrastructure buying. During the training buildout, GPU availability drove many customer decisions, even when capacity sat in a remote location. Cochrane said inference changes that pattern because applications need compute close to end users, employees and regulated data.
What is Vultr's AMD partnership?
Vultr says its AMD relationship began with EPYC CPUs and has expanded to include AMD Instinct GPUs. The company is using that CPU and GPU combination across its cloud regions as the basis for AI infrastructure it says is built for price-performance and global deployment.
Cochrane also argued that CPUs remain central to AI systems as new workloads emerge. That is a practical point for infrastructure buyers: inference services, agents and application logic do not run on GPUs alone, and cloud providers are competing on full-system efficiency rather than accelerator supply in isolation.
Open stacks and sovereignty are the go-to-market message
Vultr is positioning open composable stacks as a commercial wedge against hyperscalers. Cochrane said the company wants partners to package software and infrastructure through the Vultr marketplace, rather than push customers into a proprietary stack controlled by a single cloud provider.
At AMD Advancing AI, Vultr announced a joint solution with VAST Data, SUSE and AMD for robotics, financial services and healthcare use cases. According to Cochrane, the offering is intended to be deployed from Vultr’s marketplace in about 30 seconds. The company did not disclose customer names, pricing or adoption figures for that package.
Sovereign AI is the other durable theme in Vultr’s pitch. Cochrane described national AI infrastructure as comparable to telecoms or water treatment, meaning a basic capability that countries will expect to control. Vultr says its compliance work across 33 regions is part of its advantage as governments and regulated industries require local infrastructure options.
The positioning is clear enough: Vultr wants to be seen as an AI infrastructure specialist rather than another generic cloud provider. The harder test will be whether enterprises with five-year capacity plans move meaningful inference workloads away from hyperscalers, and whether AMD-based stacks can keep meeting Vultr’s stated cost and performance claims as agentic AI demand grows.
Cochrane said Vultr is scaling through enterprise contracts and suggested the company may make future announcements on capacity and financing. No timing, amount or counterparty details were disclosed.
This story draws on original reporting from SiliconANGLE.
