Deno Is Over (Cloudflare Shut It Down)

Author: Better Stack · Watch on YouTube

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: Deno Is Over (Cloudflare Shut It Down) · Channel: Better Stack · 04:08 · Watch on YouTube

Topics: AI agents, Software engineering with AI, AI infrastructure and hardware, Open source and software ecosystems

Summary

Cloudflare is bringing the Deno team aboard, but Deno says it will stop developing the Deno runtime after one year, while Deno Deploy will shut down after six months. The video argues that Cloudflare’s main interest is the team and Seldi, a project for running scalable Durable Objects on your own servers, which could be merged into the open-source Workers runtime. Cloudflare says self-hosting offers customers an exit option that can make them more willing to adopt Workers. The deal may benefit Cloudflare’s Workers model, but leaves Deno users facing an uncertain future for a runtime the host says people genuinely liked.

Key points

Deno will support its runtime with monthly bug-fix and security releases for one year, then end its development; it will remain open source.
Deno Deploy will shut down after six months, while JSR will continue on Cloudflare infrastructure.
Rusty V8 will remain supported, with work underway to bring it into Worker D.
Cloudflare’s acquisition is presented as being primarily about the team and Seldi, not the Deno runtime.
Durable Objects are small servers with their own SQLite databases, addressed by name; the video uses a chat channel as an example.
Seldi is described as a Rust binary and storage bucket that lets Durable Objects scale on self-hosted servers.
Ryan and Burt will lead efforts to make self-hosting Workers a first-class option, with Seldi potentially merged into Worker D.
Cloudflare argues that an open-source runtime and an escape hatch can help win large customers, rather than simply locking them in.
The video notes that Supabase Edge Functions and Netlify Edge Functions use Deno-compatible or Deno-based runtimes, and says their response is not yet known.
The host sees self-hosted Durable Objects as a significant development for Workers, but says Deno’s runtime is now on a one-year clock.

Chapters

Start	Chapter	Summary
00:00	Deno’s timeline and product changes	The video explains the runtime’s one-year support period and outlines what will happen to Deno Deploy, JSR, and Rusty V8.
01:01	Why Cloudflare wants the Deno team	The host introduces Durable Objects and explains how Seldi aims to make them scale on self-hosted infrastructure.
02:35	Self-hosting and the lock-in argument	Cloudflare’s open-source approach is presented as an escape hatch that can encourage large customers to adopt Workers.
03:38	What the deal means for Deno and Workers	The host weighs the potential upside for Workers against the end of Deno runtime development and its impact on users.

Quotes

Deno is over. — 00:00

After that year, we will end our development of the Deno runtime. — 00:31

Embarrassingly, it didn’t work. — 01:32

the escape hatch is what gets big customers to sign up in the first place. — 02:35

Self-hosted durable objects with Cloudflare behind them is actually a pretty big deal. — 03:38

Concepts and entities

Deno, Cloudflare, Deno Deploy, JSR, Rusty V8, Durable Objects, workerd, Shopify, Supabase Edge Functions, Netlify Edge Functions, Bun, Anthropic

Full transcript (original language)

00:00 Deno is joining Cloudflare, but a year from now, the Deno team is going to stop developing the Deno runtime. Deno is over. So, why did Cloudflare buy them? >> [music] >> So, here’s the deal. It’s just been announced that the entire Deno team is joining Cloudflare. But, if you scroll down a bit in the Deno blog post, you will find, “We will support the Deno runtime for another year with monthly releases containing bug fixes and security updates. After that year, we will end our development of the Deno runtime.” “Deno will remain open source,

00:31 and we welcome others who want to continue its development.” You won’t actually find this mentioned in the Cloudflare post about this, by the way. You have to read the Deno one to know that this is going on. As for the other products of Deno, Deno Deploy, that’s serverless hosting, that gets 6 months and then it shuts down completely. JSR, the package registry, keeps running and moves to Cloudflare infrastructure. And Rusty V8 stays supported, and they’re working on bringing that into Worker D. So, the question is, why are Cloudflare buying Deno and then ditching the runtime? Well, the short answer is, they aren’t buying it for the runtime. They

01:01 want the team and a project that Deno shipped in August called Cell D. To see why, you need to know a bit about Durable Objects. A Durable Object is pretty much a tiny server with its own SQLite database, and you address it by name. Ryan’s example is a chat app with one Durable Object per channel. Every channel gets its own little server and its own data. So, more channels just means more objects spread across more machines. The scaling is built into how you write the app. And that’s something that Ryan’s actually wanted ever since he made Node. His original Node.js demo was a chat server, and he says it’s always bothered him it ran on one server

01:32 on one thread. The big catch with Durable Objects, though, was they only really ran on Cloudflare. Worker D, which is the open source Workers runtime, only runs them as a single instance, which is fine for testing, but doesn’t really scale well. And Cloudflare knew this. Kenton Varda, who wrote the Cloudflare half of this post, says there’s been to-do comments about this ever since its original release. And he even tried to build it himself last spring, and in his words, “Embarrassingly, it didn’t work.” So, Seldi is Dino’s answer to all of this. Durable objects that scale on your own machine. It’s one Rust binary plus a

02:04 storage bucket and it runs your existing Cloudflare app. Durable objects are all on your own servers. So, Dino had actually built the thing that Cloudflare hadn’t managed to, which is why they bought them in. Now, Seldi can get merged into Worker D and Ryan and Burt will lead the effort to make self-hosting workers a first-class option. For Ryan, he says this is the same path he’s been on. Dino, then Dino Deploy, then Seldi. Running Deploy showed him how messy cloud infrastructure gets and Seldi was his attempt at the opposite. The runtime is just no longer part of his path anymore. Now, one question you might have after

02:35 all of this is why Cloudflare wants it to be easier to self-host Cloudflare. It’s a pretty common claim that Workers is built to lock you in and Seldi lets you walk away. Well, Kenton says that that theory is wrong. When they pitched Workers for Platforms to Shopify in 2022, Shopify wouldn’t build on it unless the runtime was open source. So, they open sourced Worker D and it’s the same code they run in production. Some customers have even used it to leave. His point is that the escape hatch is what gets big customers to sign up in the first place. It’s also worth mentioning that since this is 2026, of

03:07 course, there’s an AI angle to this, too. Ryan says that durable objects are particularly useful for agent harnesses with cheap serverless execution, persistent state and websockets and that’s why Seldi focuses on them. So, if you’ve built on Dino, you now have a one-year clock and that’s not even just people running Dino directly. Supabase Edge Functions run on a Dino-compatible runtime and Netlify’s Edge Functions run on a runtime based on Dino. I haven’t seen either company respond yet, but I’m going to guess they’ve got some hard decisions to make. I also think this is part of the bigger picture going on at the moment. Bun joined Anthropic last

03:38 December. OpenAI announced that it was buying Astro, the people behind UV and Ruff, and now Dino is joining Cloudflare. The difference with those two, though, is that Bun and UV are carrying on, but Dino isn’t. So, is this a good thing overall? For the Workers model, I think yes. Self-hosted durable objects with Cloudflare behind them is actually a pretty big deal. It takes most of the lock-in argument off the table. But for Deno, it’s a one-year clock on a runtime that people genuinely liked, and I think a lot of people are going to be sad, but Deno is now over. Let me know your thoughts in the comments down below. While you’re there,

04:08 subscribe, and as always, see you in the next one.