Google Cloud unveils Agent Compute Engine, purpose-built compute for AI agents

Summary

At AGNTCon + MCPCon on October 22, Google Cloud introduced Agent Compute Engine, a compute service that gives every AI agent its own hardware-isolated microVM. When an agent waits on a person or an external system, the microVM pauses and later resumes from a saved state. Credentials are kept out of the microVM and injected at egress. According to the company, the service already powers Gemini's work-task agents, with a customer preview coming in the next few months.

Why it matters for our work

It shows that as agents scale up, cost and security become infrastructure problems. Parking idle agents while they wait points to a way to cut operating costs for approval-heavy, long-running workflows.

Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.