Vultr Selects HPE and Nvidia for AI Inference Infrastructure

Cloud infrastructure provider Vultr has chosen Hewlett Packard Enterprise and Nvidia to power its next generation of AI data centers, marking a strategic bet on enterprise demand moving from AI experimentation to production deployment.

The partnership, announced at HPE Discover 2026, will see Vultr deploy Nvidia GB300 NVL72 systems through HPE's AI Computing portfolio. The infrastructure will incorporate Nvidia Spectrum-X Ethernet networking and HPE liquid cooling technologies. Neither company disclosed financial terms, deployment schedules, or facility locations.

Why it matters

The deal reflects a fundamental shift in AI infrastructure economics. As organizations move beyond proof-of-concept projects to production systems that affect operating margins and staffing decisions, demand is pivoting from raw training power to inference throughput and cost efficiency. This transition is creating new opportunities for cloud providers who can deliver capacity faster than traditional enterprise procurement cycles allow.

From Training to Production

Vultr CEO J.J. Kardwell told reporters that customer demand patterns have transformed over the past year. Three years ago, most GPU requests came from startups training new models. Today, a growing portion comes from organizations running production AI services tied to customer-facing applications and core business operations.

"When you reach a point in the market where you're seeing operating margins for public companies be favorably impacted, we're now seeing changes to the way large public firms think about staffing levels because of the economic impact of these AI capabilities," Kardwell said during a media briefing.

Ron Westfall, vice president and practice lead for networking and infrastructure at HyperFrame Research, characterized the moment as an inflection point. "The industry is shifting its spending focus toward infrastructure optimized for production throughput, cost efficiency and rack-to-rack networking rather than just raw training power," he explained.

Networking Becomes Critical

As AI deployments scale beyond individual servers, networking performance emerges as a bottleneck. Kardwell noted that constraints appear quickly once workloads extend across racks, making high-bandwidth fabrics essential to cluster design.

"Very quickly, the bottleneck is the second you leave that rack," Kardwell said, highlighting the importance of east-west traffic management in large AI clusters.

Future Vultr deployments will feature Nvidia Spectrum-X networking with 400 GbE and 800 GbE connectivity designed for rack-scale GPU architectures. Westfall emphasized that optimized network architectures are becoming a competitive differentiator as cloud providers compete on stable throughput and job execution speed.

Hybrid Infrastructure Strategies

Kardwell said enterprises are abandoning either-or infrastructure decisions in favor of hybrid approaches. "The days of companies having a strategy that's just all on-prem or all cloud are over," he noted. Large organizations continue building their own AI infrastructure while simultaneously using cloud providers to access the latest GPU platforms and additional capacity more quickly.

Traditional enterprise procurement cycles, which typically span six to eighteen months, struggle to keep pace with AI infrastructure evolution. By the time purchasing decisions are finalized, desired capacity is often unavailable, creating demand for cloud-based alternatives.

Vultr operates cloud infrastructure across 33 data center locations in 17 countries. Kardwell said data sovereignty requirements are becoming more important as enterprises integrate AI systems into core business processes, making geographic distribution a strategic asset.

HPE CEO Antonio Neri said Vultr's selection validates the importance of purpose-built AI data center architectures. "Vultr represents a new generation of AI cloud providers," Neri stated.

These details were first reported by Data Center Knowledge.