Agent Substrate: Advancing Kubernetes for Agents at Scale

<img width="768" height="432" src="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-768x432.jpg" class="webfeedsFeaturedVisual wp-post-image" alt="" style="display: block; margin-bottom: 5px; clear:both;max-width: 100%;" link_thumbnail="" decoding="async" srcset="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-768x432.jpg 768w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-179x101.jpg 179w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-300x169.jpg 300w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-625x352.jpg 625w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-1536x864.jpg 1536w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-645x363.jpg 645w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-660x370.jpg 660w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-500x281.jpg 500w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-160x90.jpg 160w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-362x204.jpg 362w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-196x110.jpg 196w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-1024x576.jpg 1024w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-960x540.jpg 960w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5.webp 1920w" sizes="(max-width: 768px) 100vw, 768px" title="image2" />AI agents spend most of their time idle, waiting on a human prompt, a tool result or a model response. Kubernetes wasn’t built for that. Running one Pod per...<img width="768" height="432" src="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-768x432.jpg" class="webfeedsFeaturedVisual wp-post-image" alt="" style="display: block; margin-bottom: 5px; clear:both;max-width: 100%;" link_thumbnail="" decoding="async" srcset="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-768x432.jpg 768w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-179x101.jpg 179w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-300x169.jpg 300w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-625x352.jpg 625w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-1536x864.jpg 1536w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-645x363.jpg 645w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-660x370.jpg 660w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-500x281.jpg 500w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-160x90.jpg 160w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-362x204.jpg 362w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-196x110.jpg 196w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-1024x576.jpg 1024w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5-960x540.jpg 960w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image2-5.webp 1920w" sizes="(max-width: 768px) 100vw, 768px" title="image2" /><p>AI agents spend most of their time idle, waiting on a human prompt, a tool result or a model response. Kubernetes wasn’t built for that. Running one Pod per session holds memory for sessions that aren’t doing anything, and deleting Pods between turns pays a multi-second cold start every time and loses the agent’s state. RL post-training hits the same wall. Agent Substrate, now donated to the…</p> <p><a href="https://developer.nvidia.com/blog/agent-substrate-advancing-kubernetes-for-agents-at-scale/" rel="nofollow" data-wpel-link="internal" target="_self">Source</a></p>