NVIDIA 推出 Agent Substrate，为大规模智能体优化 Kubernetes
Agent Substrate: Advancing Kubernetes for Agents at Scale
NVIDIA 发布 Agent Substrate，用于解决 Kubernetes 运行 AI 智能体时的资源浪费问题。原文指出智能体大部分时间在等待人类提示、工具结果或模型响应，按会话保留 Pod 会为空闲会话占用内存，而每轮之间删除 Pod 又要付出数秒冷启动并丢失智能体状态，RL 后训练也遇到同样瓶颈。Agent Substrate 已捐赠给相关组织。
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...
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…
来源：NVIDIA Technical Blog · developer.nvidia.com