The team behind Deno is joining Cloudflare, and the move comes with an expiration date for the products developers know it by. In an announcement on the Deno blog on Friday, Ryan Dahl said the entire Deno team will move to Cloudflare and put its future work into a shared platform, rather than continuing to develop a separate JavaScript runtime and hosting service. The companies did not disclose financial terms.
Dahl created Node.js before starting Deno as a rethink of how server-side JavaScript should work, with a stronger security model and a complete built-in toolchain. That project is now winding down on a published schedule — and for developers building on it, the schedule is the most important part of the announcement.
Two deadlines to plan around
The Deno runtime itself gets one more year of support, with monthly releases carrying bug fixes and security updates. After that year, the team ends its own development of the runtime. Deno remains open source, and the announcement explicitly invites others to carry its development forward.
Deno Deploy, the company’s hosting service, has a shorter runway: it will keep operating for six months before shutting down. Paying customers will get migration support for moving to Cloudflare Workers, according to the announcement. Two pieces survive the transition intact. JSR, the package registry, will continue operating, with its infrastructure moving to Cloudflare. Support for rusty_v8, the Rust bindings to the V8 engine, also continues, with a stated goal of integrating it into workerd, Cloudflare’s open-source Workers runtime.
Why Cloudflare wanted this team
The logic of the deal sits in a project the Deno team shipped in August. Called celld, it is an open-source implementation of Cloudflare’s Workers programming model and Durable Objects, designed from the start to be self-hosted and to scale on a developer’s own infrastructure. In a joint post on the Cloudflare blog, Dahl and Cloudflare engineer Kenton Varda frame celld as the bridge between the two companies’ work.
Cloudflare is unusually candid about the gap celld fills. Workerd — the same runtime code Cloudflare runs in production, released as open source — implements Durable Objects only in a single-instance way that the company itself describes as good enough for local testing but unable to scale. Its production system for routing Durable Objects was built for hundreds of locations and leans on external services maintained by site reliability engineers, which makes it a poor fit for self-hosters. Cloudflare also concedes that it has not built out the tooling and surrounding services that would make self-hosting workerd practical, and that only a few people run it in production today.
What gets built next
Dahl and Bert Belder will lead a new effort to make self-hosting workerd a first-class, supported way to run applications built on the Workers model, merging code and ideas from celld back into workerd. The stated ambition is broad: the combined teams want that programming model to become the default way to build servers, whether the code runs on Cloudflare’s network or on infrastructure the developer controls. Cloudflare says more announcements about the work will arrive in the coming months, and developers who do not want to wait can already self-host celld or workerd.
AI is part of the pitch. Dahl argues that Durable Objects bring together properties that are especially useful for agent harnesses — inexpensive serverless execution, persistent state, WebSockets, and a high-level JavaScript interface — at a moment when, as recent episodes with AI agents acting on live systems have shown, the infrastructure underneath agents is becoming a product category of its own.
What to watch
Three things will determine how smooth this landing is. First, the migration path: Deploy customers have six months, and the details of the promised migration support will matter more than the headline. Second, whether a community maintainer steps up for the Deno runtime before the team’s final year runs out. And third, the first concrete releases from the merged workerd and celld effort — the test of whether “run Workers anywhere” becomes a supported product or remains a promising repository.
