Vercel is running 116 deployments a second, and rebuilt its CDN metadata for the cold path

Guillermo Rauch said on September 10 that about 10 million deployments are made on Vercel every day, and 2.35 billion in total since the platform started. Do the division. That is 116 deployments a second, and it means Vercel's entire historical deployment count would be recreated in 235 days at the current rate.

Guillermo Rauch@rauchg

Each day ~10 million deployments are made on Vercel, with 2.35 billion made to date.

Vercel is one of the most heavily multi-tenant systems in the world. Billions of application deployments co-exist and are accessible ('routable') on our CDN at any given moment.

Underlying our CDN is a global metadata store that synchronizes within hundreds of milliseconds, globally. e.g: when you roll back, change config, add routes, etc.

We just made this system 91% faster at p99, and in the process sped up the build→deploy pipeline. All of this while the system is under immense pressure from the growth in agentic deployments.

Great read on the internals of Vercel from our CDN engineering team:

Rauch names the cause in his own post, in a clause most people scrolled past: all of this is happening "while the system is under immense pressure from the growth in agentic deployments". A deployment used to be a decision somebody made. Now it is a side effect of a loop.

If you ship on Vercel, you are probably no longer the thing generating most of that traffic.

So what does a platform do when the button gets pressed by a program? The engineering post Vercel published the same day is the more interesting document, and the answer is in it.

Every request to Vercel passes through a CDN that executes, on average, more than 80 million routing instructions per second. Part of that work is looking up which paths exist and how to serve them. Vercel used to store that metadata as a separate object per path, fetched and cached individually — fine for a small site that deploys twice a month, and quietly terrible for a large one that deploys constantly, because every new deployment mints a fresh set of cache keys and the first lookup for each path misses.

And that is probably the agent tax, stated precisely. Agents do not make your site bigger. They make it redeploy before the cache is warm.

The fix is a good piece of plumbing. Vercel grouped path metadata into bounded shards of roughly 200 KB, each holding sorted key-value JSONL records with an inline index of fixed-width Base64 pointers, so the routing process can binary-search the encoded paths and then parse only the one line it needs. One fetch warms many paths. They tried multi-megabyte shards first, found the per-process LRU cache missed too often (requests scatter across processes in a region), and came back down.

But look at what did not move. The median stayed around 0.7 ms. This was never a speed project. It was a variance project, and the standard deviation falling from 44.9 ms to 19.0 ms is the honest headline. Most requests were always fast.

The nasty ones were the cold ones, roughly speaking, and cold is what a deploy-every-minute workload produces.

Our read: the 10-million figure is a load statistic dressed as a growth statistic, and the companies that will feel this first are the platforms whose pricing assumes a human rhythm. Build minutes, preview environments, seat-based plans — all of them were priced when a deploy meant a person had finished something. Vercel's own answer so far is to make the cold path cheap rather than to reprice it, and it took 16.6 seconds out of the deploy pipeline while it was there (about 10% of the deploy step, closer to 25% for metadata-heavy projects). We would expect at least one major CI or preview-environment vendor to introduce explicit agent-rate pricing before the end of the first quarter of 2027. If instead the industry absorbs it quietly, the marginal cost of a deployment is lower than we think, and we would want to see someone's margins to believe it.

One caveat, and it matters: "deployment" is Vercel's own word, counted by Vercel, with no definition published. Preview deployments on every push are almost certainly in there, which is exactly where agent traffic lives. The number is real and the denominator is unstated.

The detail we enjoyed most is buried in the rollout. Running both lookup paths in shadow mode for weeks turned up one disagreement: an old encoding packed paths into RFC 2047 encoded words, and it broke when an emoji was split across two of them. Somewhere there is a URL with a cat face in it that has been quietly wrong for years, and nobody noticed until a migration went looking.