The laptop just stopped being the runtime for agentic coding. Anthropic took Claude Code cloud sessions out of research preview this week. A cloud session runs on Anthropic managed infrastructure, or your own self hosted environment, and keeps working after you close your laptop. Start one from claude.ai/code, the Code tab in the mobile app, the desktop app, or claude --cloud in the CLI. It comes paired with a redesigned Projects feature: a project is one ongoing conversation where Claude acts as coordinator, decides what becomes a thread, and each thread is its own full cloud session running on its own branch and its own copy of the repo. Existing subscribers get a one time credit to try it, 100 dollars on Pro, 250 on Max, separate from normal usage limits. Every major coding tool is converging on this same pattern right now. Cursor shipped Cloud Agents running in isolated dev environments. OpenAI added parallel cloud sandboxes to Codex. Cognition's Devin already runs as managed cloud VMs by default, merging 67 percent of its PRs. GitHub has been pushing Copilot workspaces the same direction. The laptop as the place code actually executes is becoming the exception, not the default. I have seen this arc before. CI moved from running tests on someone's laptop to hosted runners. Local dev environments moved from a Vagrant box under your desk to Codespaces and Gitpod. Every time, compute detaches from the person and execution becomes a service you request instead of a process you babysit. The part nobody has fully solved yet is not where the agent runs, it is knowing what a background agent did while you were not watching and whether it burned your quota doing it. That is exactly why monitoring tools like Cursor's new Rollouts feature matter as much as the cloud runtime itself. My take is that cloud execution is becoming table stakes across the category this year. The real competition is shifting to what wraps around the session: planning, verification, and cost control. nitter.cf/ClaudeDevs/status/2102…