What changed
Google announced Agent Compute Engine, a service that provides each AI agent with its own hardware-isolated micro virtual machine (microVM). The service already powers Gemini and is set to open for preview in the coming months. Each agent runs inside a microVM secured by Google's Titanium chip. When an agent is idle, its VM pauses automatically, freeing up the machine for other tasks. Shared memory between VMs reduces the cost of running each one.
What this means in practice
The service aims to improve efficiency in agentic systems, where VMs are a significant cost factor. Google claims it can fit ten times as many agents on a single server. In a test with 200,000 agents, suspend and resume operations took less than 500 milliseconds. The service also includes token tracking, model routing, and support for third-party identity providers like Ping Identity and SailPoint. These controls will extend to Cloud Run, GKE, and managed endpoints.
Limitations and availability
The service is currently in preview and will be available for testing in the coming months. The supplied announcement does not specify exact availability dates or pricing details. The use of Google's Titanium chip for securing microVMs is highlighted, but no further details on its capabilities or limitations are provided. The service is part of the open-sourced Agent Substrate, a Kubernetes runtime, but no additional information on its features or integration is given.
