SUSE Rancher and Vultr want to break AI infrastructure free from the hyperscalers

Host A: Welcome to DevTools Radio, I'm your host, and today we're talking about something that a lot of platform engineers have strong feelings about — breaking free from hyperscaler lock-in when it comes to AI infrastructure.

Host B: Oh, and "strong feelings" might be an understatement. Cloud bills for AI workloads have been absolutely brutal, especially when you're running heavy inference pipelines on Kubernetes.

Host A: Right, so here's the news — SUSE Rancher Prime and SUSE AI have joined the Vultr Marketplace, and the people behind this deal are pitching it as more than just a partnership. They're calling it a blueprint for sovereign, open-source AI infrastructure.

Host B: Okay, I love the ambition there. But let's break that down — what does "sovereign AI infrastructure" actually mean in practice for, say, a DevOps team trying to ship something?

Host A: Great question. Essentially, it means you're not locked into AWS, Azure, or Google Cloud to run your AI inference workloads. Vultr brings the GPU horsepower — we're talking B200s, H100s, MI300X instances across 32 global regions — and SUSE Rancher handles the Kubernetes cluster management and security layer.

Host B: So it's kind of sitting in this interesting middle ground — not a full hyperscaler, but also not a "roll your own Kubernetes on bare metal in your basement" situation either.

Host A: Exactly, and that's actually the pitch. Vultr's CMO Kevin Cochrane described it at KubeCon Europe as offering freedom, choice, and flexibility — and he was pretty pointed about the alternatives, including what he called "neo-clouds."

Host B: Neo-clouds — I haven't heard that term before. What's the concern there?

Host A: So these are the well-funded AI infrastructure startups that have raised massive capital to offer raw GPU power. Cochrane's warning was pretty direct — he said enterprises end up walking away because when the CISO, the SecOps team, and the network team show up with their compliance checklist, there's just not a lot there to satisfy them.

Host B: That tracks. Raw GPU access is great until your security team asks about data sovereignty, compliance frameworks, and zero-trust networking — and you get a blank stare back.

Host A: And that's where SUSE's stack fills the gap. You get Rancher for cluster orchestration, SUSE AI for inference and training, and built-in zero-trust security. Cochrane's argument is that Vultr has been a mature public cloud for 14 years — it's not a startup experiment.

Host B: The timing also makes sense. Cochrane noted that true enterprise adoption of AI for mission-critical systems hasn't fully landed yet, but the shift is happening now, driven by enterprise inference becoming a real priority for platform engineering teams.

Host A: And for those teams, the message is essentially — take the cloud-native principles you've already established, and extend them to AI-native apps without handing your entire budget and your architecture to a hyperscaler.

Host B: Honestly, if the pricing is as competitive as they're claiming and the compliance story holds up under scrutiny, this is a conversation a lot of CTOs are going to want to have soon.

Host A: Absolutely. It's a space worth watching closely. That's going to do it for today's episode of DevTools Radio — thanks for tuning in.

Host B: Stay curious, keep building, and maybe go check that cloud bill one more time. We'll catch you next time.