On Friday, October 9, Cloudflare announced it is absorbing the entire Deno team. The Deno runtime gets one more year of monthly bug-fix and security releases, then development ends. Deno Deploy shuts down after six months, with migration support for paying customers. JSR survives. No purchase price appeared in the announcement, which is itself a data point.

Ryan Dahl wrote Node.js, spent eight years building the runtime meant to correct Node.js, and then — this is the part getting lost — spent 2026 building an open-source, self-hosted clone of Cloudflare Workers. Deno’s celld project is a Rust daemon that embeds V8 and runs Workers and Durable Objects on hardware you own: no control plane, nodes coordinating through a single S3-compatible bucket, each cell its own SQLite database replicated to object storage. One teardown circulating on YouTube pegged the cost advantage at 85x. I cannot verify 85x and neither can you, but the direction of that claim is not seriously contested.

So as of Friday morning, Cloudflare faced a real competitor offering a cheaper, portable, self-hosted substitute for its core product. By Friday afternoon, that competitor’s authors worked for Cloudflare. The New Stack had the story up before the close of business.

The Contestability Argument Just Got a Term Sheet

For fifteen years, the best argument against antitrust action against large platforms has been contestability. Don’t sue them; out-build them. If a rival shows up with a better product at a lower price, customers move, discipline gets applied, and nobody needs a regulator. It is a good argument — genuinely one of the better ones in the policy toolkit — and Friday was its field test.

Deno was not a hobby project. It raised $21 million in June 2022, per TechCrunch, with Sequoia, the Mozilla Corporation, Shasta Ventures, Long Journey Ventures, Rauch Capital and Four Rivers Ventures on the cap table. It had a decade of credibility from Node.js and a technically serious alternative to Workers. And when the alternative got good enough to matter, the incumbent’s cheapest, fastest, most legally unassailable response was not to compete on price. It was to buy the team at whatever the secondary market charged and let the runtime expire on a one-year clock.

Nobody is going to jail for that. That is the point.

An Open-Source License Is Not a Maintenance Contract

Here is where my own side should start squirming. The reflex on the right is that open source plus a permissive license equals permanent contestability — the code cannot be un-shipped, so the check survives any acquisition. Deno will remain open source, we are told, and others are welcome to continue development. celld is open source too.

Try running a payments system on a repository whose last funded maintainer left for Cloudflare in October.

The license survived. The thing did not. What Cloudflare bought for an undisclosed sum was not a runtime or a daemon — both of which it can clone at leisure — but the roadmap, the judgment, and the small group of people who know why the hard parts are hard. No license text obliges a company to keep paying salaries to maintain a competitor’s escape hatch.

And remember the interop push between Node.js, Deno and Cloudflare to move applications among the three without a rewrite. That portability promise was arguably the strongest competitive constraint on Workers pricing. One of the three participants now works for another.

Nobody Broke a Law, Which Is the Problem

The left’s preferred remedy has the same exposure. Break them up, mandate interoperability, force an open API — fine. But a mandated interface is a piece of paper, and the people who make it do something are employees with mortgages. They are for sale, and the buyer with the most to gain from their departure is the incumbent they were constraining. A mandate that does not fund a maintainer is a museum placard.

A junior platform engineer at a payments company who spent the summer moving workloads onto celld, writing in a public developer Discord, put it better than I will: “We didn’t switch to celld to save money. We switched because it was the only version where we could read the code and leave. Turns out you can’t leave when the people who wrote it leave first.”

That is not a scandal. It is not even surprising. It is the exit market functioning exactly as designed — allocating scarce engineering talent to whoever values it most, which is almost always whoever has the most to lose from it. The uncomfortable implication for anyone who prefers markets to regulators is that the market’s answer to platform power is a signing bonus.

So watch three things. Watch whether anyone credible forks the runtime without a payroll behind it. Watch JSR, which survives for now and whose survival is entirely at Cloudflare’s discretion. Watch whether Workers pricing moves over the next eighteen months.

My prediction: it does not move, because it does not have to. The competitor did not lose. It got hired.

Sources