I was halfway through a second coffee on Friday when I saw the news scroll by: Cloudflare bought Deno. Not "invested in," not "partnered with", bought it outright, announced October 9th, terms undisclosed like always. My first reaction wasn't shock. It was that specific tired feeling you get when a pattern completes itself for the hundredth time.
Quick background for anyone who wasn't following server-side JS drama a few years back: Ryan Dahl created Node.js, then left the project, then gave a talk called "10 Things I Regret About Node.js," and then built Deno as the do-over. Secure by default, TypeScript out of the box, no node_modules nightmare. I actually tried it for a side project back in 2020 and liked a lot of it, even if half my dependencies didn't work yet. Deno raised money too, $26 million total, with a Sequoia-led Series A at some point along the way, and it built Deno Deploy as its hosted product, going head to head with Cloudflare Workers and Vercel.
And now the company goes to the thing it was built to compete with.
Here's the part that actually matters if you have anything running on this stuff: Deno Deploy shuts down within six months. Cloudflare says it'll offer migration help to paying customers, which is nice, but six months is not a long runway if you've got production traffic sitting on somebody else's edge network and you weren't planning your Tuesday around a sudden migration project. The open-source runtime itself gets a slightly longer goodbye, a year of bug fixes and security patches, and then Cloudflare is done with it. It stays open source, they say, and "welcomes the community" to keep it alive, which is corporate-speak for "someone else can deal with this now."
What Cloudflare actually wanted, from what I can tell, isn't the runtime people install on their laptops. It's something called celld, Deno's open-source reimplementation of the Workers programming model, which Cloudflare plans to fold into workerd so companies can self-host Workers-style apps on their own servers. Ryan Dahl and Bert Belder are moving over to lead that effort. So the guy who quit Node to build something better than Node is now going to go make Cloudflare's proprietary-ish runtime more self-hostable. There's a version of this that's a genuinely interesting technical move. There's also a version of this that's just acquihire-and-sunset with better PR framing, and I honestly can't tell you which one it is yet, and neither can anyone writing about it right now, no matter how confident the headlines sound.
I'll say the unpopular-sounding thing: I don't think this is really about Deno at all. It's about Cloudflare wanting a credible answer to "what if I don't want to be locked into your cloud forever," because that question has gotten louder this year, and buying the team that already built half the answer is cheaper than building it from scratch. Fine. Makes sense from a boardroom chair. It just means Deno-the-product, the thing a bunch of people bet side projects and in a few cases actual companies on, was never really the point. It was a stepping stone wearing a product's clothes.
The reaction I've seen from developers has mostly been annoyance, not surprise, which tells you something. Nobody's shocked anymore when a scrappy runtime or a beloved dev tool gets folded into a platform and quietly wound down over a fiscal year or two. We've all watched this movie. I lived through Google killing products I liked enough times that I stopped being precious about any tool that isn't either boring-old-and-stable or something I run myself with no hosted dependency. That's maybe the actual lesson here, if there has to be one: anything with a "Deploy" button attached to a startup's name is a bet, not a foundation. Build accordingly.
Anyway. If you've got something running on Deno Deploy right now, go start that migration conversation before the six months turns into five, then three, then an email you didn't read carefully enough.