Vercel adds always-on workers to run more of your backend
The New Stack Frederic Lardinois
Vercel just added always-on background workers, plus GA Python support and Drives. It’s a push to host more of a backend, not just the frontend.
Based on reporting by The New Stack, Frederic Lardinois — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Vercel is trying to pull more of the backend onto its own platform. At Ship in San Francisco on Thursday, it introduced Daemons, a service for background workers that stay alive continuously, while also making Python support generally available. Drives, its storage system for AI agents, is now generally available too.
Daemons is in public beta, and it is aimed at a very specific kind of job: work that doesn’t fit neatly into request/response traffic. Developers can deploy a container, set how many instances should run in each region, and let Vercel keep them alive. If one exits or crashes, Vercel restarts it. The workers don’t wait for web requests. They pull work from a queue or database and process it as it arrives.
CTO Malte Ubl frames this as Vercel moving further down the stack. The company already has queues, durable workflows, and Services for frontends and backends together, but he says that still leaves out the sort of backend processing that runs against a queue for a long time. He points to Vercel’s own deployment pipeline, where jobs can’t just pile up in the request handler forever. GitHub rate limits, he says, are one reason to back off, and Vercel’s older compute tools did not give developers enough control over concurrency for that style of worker.
The company’s example setup mixes a Next.js frontend, a Python API, and a containerized video processor that includes Python dependencies and FFmpeg. That worker pulls jobs from Vercel Queues to generate playback files and thumbnails. Ubl says transcoding is a classic fit here: CPU-heavy work where the goal is simply to keep machines busy all the time. He puts Daemons at the opposite end of the spectrum from sandboxes, which are temporary and invoked by requests.
Vercel says 5,000 teams have made 150,000 container deployments since container support entered public beta, and 150,000 teams already run Python backends on the platform. Python support now covers FastAPI, Flask, and Django, with CPython and PyPI dependencies handled on Vercel’s side. The company also says its control plane API processes about a billion requests a month on Vercel, and that many of its asynchronous jobs are now moving onto Daemons. On top of that, Drives is meant to let AI agents keep using the same files across sandboxes, with data landing on local NVMe first and then streaming to object storage.
My take — AI-written commentary, not fact-checked reporting
This is the kind of move that makes platform lock-in look almost polite. If Vercel can host the frontend, the queue, the worker, the Python app, and the agent storage too, there’s less reason to leave and more reasons to forget who owns the plumbing. Convenience wins again, which is basically the oldest story in cloud software.
Read more about this at: The New Stack
