Cloudflare reportedly buys Deno, and the runtime gets one year to live

One announcement, one source so far

On October 9, 2026, Simon Willison flagged on his blog a piece of news that probably made a few deno.json files spill their coffee: Cloudflare is reportedly acquiring Deno outright. His post points to the announcement Cloudflare published on its own blog. As we write this, though, that’s the only source we have, so take everything below as “reportedly” until someone else confirms it.

Cloudflare wouldn’t really be after Deno itself. It would be after celld. The Deno team shipped the first version of that project in August: an open source implementation of the Durable Objects model from Cloudflare Workers. Cloudflare reportedly wants to use it to turn self-hosting workerd into a fully supported way to build and run applications with the Workers programming model.

A year of maintenance, then lights out

Here’s the part that stings. According to the announcement Willison quotes, Cloudflare would keep maintaining the Deno runtime for one more year, with one release a month limited to bug fixes and security updates. After that, Cloudflare would stop developing it. Deno would stay open source, and the company says it would gladly welcome anyone who wants to carry it forward.

If your team runs Deno in production outside Cloudflare, that means twelve months of patches, nothing else announced, and then a repository whose future rests on a community that hasn’t raised its hand yet. Now is a good time to check whether your code leans on Deno-specific features or would run elsewhere without a fuss. A migration planned over a year is a gentle stroll. The same migration forced on you by a vulnerability nobody is around to patch is a sprint in the dark.

The why, according to the creator

Ryan Dahl, who created both Node.js and Deno, explained in a Hacker News comment that the decision was made jointly and that he backs it. His assessment is blunt: Deno is well designed and carries good ideas, but it doesn’t solve big problems. The project got swallowed by Node compatibility, which forces it to behave exactly like Node. His line, in his own words: “Why reimplement Node? It works.” In his view, small gains in performance, ergonomics or security aren’t enough.

What excites him now is celld. He says the project works remarkably well relying only on object storage for coordination and persistence. He sees it as an entirely new model for server-side development, not a slightly different API for talking to the disk or the network. The irony writes itself: the runtime built to fix Node ended up having to do a pitch-perfect Node impression.

What you actually lose

Willison names the feature he’ll miss: Deno’s permission system. It lets you run a script while spelling out exactly which files and folders it can read or write, and which network hosts it can contact.

Node.js has a comparable model. It arrived with v20.0.0 in April 2023 and was declared stable in v22.13.0 in January 2025. It still can’t allow specific network hosts, though: in Node, the network is either open or closed. If you were sandboxing your scripts down to the domain name, moving to Node means trading your scalpel for an on/off switch.

Deno set out to be a better-built Node, and it now has a year to find a community willing to prove it. Everyone else can go ahead and put a pin in the migration calendar.