Back Issues/Search Home → Calendar → Archive → RSS → Subscribe → Services → Vendors → Blog → Current Issue → Popular → Newsletters (LinkedIn, Substack)

All issues › Volume 344, Issue 2 › IT Vendor News › Google

The Open AI-Native Stack on Kubernetes: From Open Models to Distributed Inference and Agents

Google, Tuesday, November 10th, 2026

Google Cloud outlines an open-source AI-native stack on Kubernetes covering RL post-training, inference and agents.

Ahead of KubeCon + CloudNativeCon North America 2026, Google Cloud describes an open-source stack for running AI workloads on Kubernetes and GKE.

It identifies hurdles such as idle accelerators during reinforcement learning, cache-blind load balancing, fragmented compute frameworks and the risk of running untrusted agent code.

The stack includes OpenRL for post-training, llm-d 1.0 for distributed inference with prefix-cache-aware routing, Ray on GKE with KubeRay, Agent Substrate for agent workflows and gVisor for sandboxing.

TorchTPU adds native PyTorch support across Cloud TPUs and GPUs in vLLM and SGLang, and the post includes sample manifests to try on clusters.

more →  ·  More from Google →