Kubernetes launched the AI Gateway Working Group. It will add standards and declarative APIs to make networking play nice with AI workloads and extend the Gateway API.
Active proposals attack two gaps. Payload processing inspects and transforms full HTTP payloads using declarative configs, ordered pipelines, and explicit failure modes. Egress gateways route outbound AI traffic out of the cluster.
This is a system shift. The group pushes AI-specific networking into the Gateway API, moving policy enforcement up to the platform so inference traffic follows standardized network rules.
