AIToday
Large Language ModelsAI Business & IndustrySiliconANGLE AIPublished: Nov 20, 2026, 10:01 JST

Vultr's Cochrane: AI agents demand full cloud, not just GPUs

Send on LINE
Vultr's Cochrane: AI agents demand full cloud, not just GPUs

3 Key Points

  1. What happened

    At KubeCon + CloudNativeCon North America 2026, Vultr CMO Kevin Cochrane said enterprises need a full cloud platform — compute, networking, storage and an open ecosystem — and predicted consolidation among neoclouds selling only bare-metal GPU capacity.

  2. Why it matters

    Cochrane argues bare-metal GPU capacity alone leaves enterprises with a second infrastructure silo, so providers that expand into full public cloud platforms may be better positioned to serve production AI workloads, per his prediction.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

Vultr's position reflects a shift in what customers ask of AI infrastructure. The company entered the cloud market in 2014 with a developer-focused platform built around performance, cost efficiency and accessibility, then expanded into AI infrastructure offering both traditional cloud services and specialized GPU resources. Cochrane described that expansion as deliberate, prioritizing enterprise workload and platform-engineering requirements over GPU availability alone, and called Vultr "the tortoise in AI infrastructure."

The requirements Cochrane outlines follow from how AI adoption is changing. Early infrastructure demand came largely from hyperscalers, AI research organizations and startups building large-scale models, often through bare-metal GPU clusters. Enterprises now want AI agents that work alongside existing systems and comply with security requirements, which Cochrane says makes complexity and a lack of specialized skills major obstacles. Vultr's answer is an integrated environment supporting more than 60 Cloud Native Computing Foundation projects that customers can deploy directly.

On data sovereignty, Cochrane framed Vultr's regional model as an isolation architecture rather than a storage question: each of its 34 global data center regions operates independently, and data within a region remains resident there. On cost, he argued some organizations will post-train open-source models and run them on infrastructure they control, while stressing that not every enterprise should build and operate its own models. He also expects platform engineers to work more closely with business units and predicted consolidation among neocloud providers that remain focused on bare-metal GPU capacity.

FAQ
Why does Cochrane say GPUs alone are not enough for enterprise AI?
He says AI agents execute code, access databases and invoke external tools, so CPUs, secure sandboxes, networking and storage must surround the GPU. Buying GPU capacity without those pieces can create another infrastructure silo.
What does Vultr's regional data isolation mean in practice?
Cochrane says each of Vultr's 34 global data center regions operates independently, so an application running in a French region should keep running even if Vultr's global control plane and other regions become unavailable.
How did one organization cut inference costs, according to Cochrane?
He described a platform engineering organization that reduced inference costs by 80% by post-training an open-source model and running it on its own infrastructure. He noted results depend on the workload, model and operational requirements.
SiliconANGLE AIRead Original Article
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleNvidia, partners build end-to-end AI ecosystem with Epson