Stateless MCP and the industrialisation of agent plumbing — the boring change that lets agents scale

The most consequential agent news of the cycle is a protocol revision that removed session affinity. It is not glamorous. Industrialisation never is — it is the moment a craft technique becomes infrastructure you no longer think about.

The Model Context Protocol's 2026-07-28 revision adopted a stateless architecture, dropping session affinity. Translated: agent backends can now scale like ordinary web services instead of pinning each conversation to a specific server. That is the difference between a demo and a fleet.

Why statelessness is the whole story

Session affinity — the requirement that every message return to the server holding its state — is what makes agent infrastructure expensive, because it defeats the load balancers and autoscalers that make normal traffic cheap. Removing it is an unglamorous operations decision dressed as a protocol one, and it is exactly the kind of change that signals a technology moving from experiment to production.

The governance shift underneath is as important. MCP originated at one lab and was handed to the Linux Foundation while it still had momentum — the rare standard given to a neutral body before a fork could fragment it. This is the first major spec change under shared stewardship, and it reads like an operator's wishlist, not a vendor's.

The plumbing metaphor is exact

You do not admire plumbing; you rely on it. As agents gain the ability to move money and coordinate across organisations, the connective tissue has to be invisible and dependable, which is precisely what statelessness buys. Betting on MCP as the substrate is now lower-risk than betting on any single model.

The agent era will not be won by the flashiest model. It will be run on plumbing nobody decided on but everyone adopted.

The Agentics — The enterprise MCP guide 2026 → · Sourcetrail — AI agents in 2026: protocol overhauls to autonomous payments →