Vultr bets on AMD and open stacks to win cloud AI infrastructure

Cloud AI infrastructure is shifting into a new cycle, defined by who can deliver the strongest performance per dollar on a global setup that also meets different countries’ sovereignty rules.

That’s the premise Vultr is executing on, according to Kevin Cochrane (pictured), chief marketing officer at Vultr Holdings LLC. The company began its AMD partnership on the CPU side with EPYC and has since expanded to the Instinct GPU line, building a full public cloud platform across 33 global regions on the back of AMD’s predictable generational roadmap. For Cochrane, the reliability of that roadmap serves as the infrastructure of Vultr’s advantage in price performance, delivering up to 33% better performance at an average of 82% lower cost than competing hyperscaler offerings.

“The CPU never went away,” Cochrane said. “It now just breathes new life because we have new types of workloads that we need to deploy.”

Cochrane spoke with theCUBE’s Dave Vellante at the AMD Advancing AI event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how Vultr is positioning itself as the enterprise cloud AI infrastructure specialist, why sovereign AI is a durable, global imperative and how open composable stacks are becoming the company’s primary path to market. (* Disclosure below.)

Cloud AI infrastructure strategy built on open composable stacks and sovereign reach

Vultr’s demarcation from the hyperscalers boils down to three compounding advantages: deep AMD partnership economics, a globally distributed footprint designed specifically for the inference era and a marketplace built around open composable stacks instead of proprietary lock-in. Cochrane said the training era produced hundreds of neocloud providers because customers needed GPU access and didn’t care where the cluster was located. The inference era is fundamentally different. Customers need low-latency, distributed compute wherever their users and employees are.

“You need to have decentralized distributed computing everywhere around the world, wherever your customers and your employees live and work,” Cochrane said. “For us, it’s not just about making capital investment decisions to build a massive new data center in one region. We need to distribute it across all 33 global data center regions.”

Sovereignty is a durable structural trend rather than a political moment, Cochrane noted. He compared national AI infrastructure to telecommunications and water treatment — a baseline that every country will eventually require. Vultr’s compliance expertise across all 33 regions has become an increasingly competitive asset. At the show, Vultr announced a joint solution with VAST Data, SUSE and AMD targeting robotics, financial services and healthcare use cases, designed as a prepackaged open composable stack that customers can deploy from the Vultr marketplace in half a minute.

“Open, composable architectures. That’s our play,” Cochrane said. “We don’t want to compete with our partners; we want to enable the ecosystem.”

Looking ahead, Cochrane said Vultr is scaling fast on the back of enterprise contracts with five-year planning horizons, and he also hinted at future capacity and financing announcements. The company’s goal is to be the cloud AI infrastructure specialist for the agentic era, delivering the most agents per watt, at the best price per dollar.

“Every single country on the globe needs critical cloud infrastructure,” Cochrane said. “We at Vultr are on a mission to make sure every single country gets a full-stack AMD CPU and GPU solution built out.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the AMD Advancing AI event:

(* Disclosure: TheCUBE is a paid media partner for the AMD Advancing AI event. Neither AMD, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE