🧩Google Donates Agent Substrate and gVisor to CNCF for Agent Infrastructure
Your agent sandboxes are about to get standardized
TL;DR
Google donated Agent Substrate, gVisor, and OpenRL to the Cloud Native Computing Foundation to tackle agent infrastructure. Agents sit idle 90-95% of the time while holding state, and Kubernetes wasn't built for that. Expect open standards for running untrusted AI code at scale.
Google just gave Agent Substrate, gVisor, and its OpenRL reinforcement learning orchestration project to the Cloud Native Computing Foundation. The goal is an open foundation for agentic applications, the way Kubernetes standardized container orchestration. If you're building agents that execute AI-generated code, this matters. Kubernetes assumes long-running apps with stable workloads. Agents create, suspend, restore, and isolate environments far more often, and that churn is expensive and risky. Substrate separates agent state (actors) from compute (workers), so a paused agent can release its resources and resume later on a pre-warmed pool. The numbers that drive this: Google estimates agents spend 90-95% of their time idle while still holding important state. gVisor adds an isolation boundary through a user-space kernel that intercepts system calls, and Substrate also supports Cloud Hypervisor microVMs for workloads needing hardware-level virtualization. Competing approaches like Kata Containers and Firecracker already exist, so the real bet here is standardization. Watch whether the CNCF ecosystem converges on this model or fragments around Kata and Firecracker.

Key Points
Google estimates agents spend 90-95% of their time idle while still holding important state, which breaks the assumptions Kubernetes was built around for long-running services.
Keep reading — it's free
Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.
Already a member? Sign in
Comments
Be the first to comment