The Rise of Neo Clouds and Self-Service AI Infrastructure

An interview with Rafay CPO Mohan Atreya, recorded by JetStor at Supercomputing 2025

This interview was conducted by JetStor at Supercomputing 2025 and features Rafay Chief Product Officer Mohan Atreya on the rapidly changing AI infrastructure landscape.

As GPU power density, cooling requirements, and cost pressures reshape the data center, enterprises and cloud providers are looking for new ways to deliver AI infrastructure with a cloud-like experience—without relying solely on hyperscalers. Mohan unpacks the rise of “neo clouds,” why self-service has become table stakes, and how Kubernetes must become critical—but effectively invisible—to the teams consuming AI platforms.

In this interview, you’ll learn:

• What “neo clouds” are and why they’re emerging now
• Why many enterprises are renting GPU capacity instead of building data centers
• How self-service changes the developer and data scientist experience
• What it means to make Kubernetes “invisible” for AI/ML workflows
• Why we’re still in the early days of AI infrastructure

Watch the full interview below to understand how AI infrastructure is being rebuilt for the next decade.

‍