Cowork moves tool execution fully to cloud, ditching local VM model
Source published ·Modelwire updated
Original coverage: Simon Willison ↗·How Modelwire adds context
The development
Anthropic's Cowork is shifting its architecture from hybrid cloud-VM execution to fully cloud-based inference and sandboxing. The original design ran model calls in the cloud but executed tool operations in a local VM shipped to users, trading capability for battery drain and session persistence issues. The redesign eliminates local compute entirely, isolating each session in its own cloud sandbox while maintaining selective device access for user files. This reflects a broader tension in AI tooling: balancing user control and data locality against operational simplicity and continuous availability.
Modelwire’s AI-generated summary of coverage from Simon Willison.
Modelwire analysis
Analyst takeOur AI-generated reading of the wider context and the next developments to watch.
Anthropic is abandoning the hybrid model not for technical elegance but operational necessity. Local VMs created friction (battery drain, session loss) that cloud-native competitors like OpenAI's Codex Cloud have already solved, forcing Anthropic to follow rather than lead on infrastructure maturity.
This tracks directly with OpenAI's Codex Cloud redesign from late September, which moved to persistent cloud environments across devices. Both moves signal that the competitive bar for AI tooling has shifted from 'works' to 'works everywhere without friction.' More critically, Cowork's redesign arrives as Anthropic faces acute customer concentration risk (Microsoft and Meta both halved Claude deployments in early October), making infrastructure reliability a business survival metric rather than a feature. Anthropic's simultaneous $100 million engineer training investment suggests the company is betting that operational excellence and developer experience can offset its loss of enterprise anchor customers.
If Anthropic ships selective device access (file sync, local peripherals) within the cloud sandbox before Q1 2027, it signals they're matching OpenAI's feature parity. If adoption metrics for Cowork don't recover within 60 days of launch, it indicates the architecture fix alone can't overcome the customer concentration problem that the October Meta/Microsoft pullback exposed.
This interpretation is generated from the summary above and the archive coverage cited below. Our methodology · Report an error
Coverage behind this analysis
These archive entries ground the connection in our analysis. They are ordered by source publication date, with links to our coverage and the original sources.
·OpenAI (YouTube)
OpenAI launches Codex Cloud for cross-device development workflows
OpenAI is expanding Codex into a cloud-native platform that persists development environments across devices, enabling engineers to start tasks on desktop and resume on mobile without context loss. The move signals a shift toward treating AI-assisted coding as a stateful, team-oriented workflow rather than isolated sessions. By bundling repository access, networking, and secret management into…
MentionsAnthropic · Cowork · Claude · Felix Rieseberg · Simon Willison
How this coverage is produced
Modelwire uses AI to generate summaries and context from source headlines, snippets, and selected archive coverage. Automated checks do not verify every claim, and items are not routinely reviewed by a person before publication. Zacaria Solis operates the site. Read the linked source for the full evidence and report errors through our corrections process.
Modelwire summarizes, we don’t republish. Simon Willison originally reported this story as “Quoting Felix Rieseberg”. The full content lives on simonwillison.net. If you’re a publisher and want a different summarization policy for your work, see our takedown page.