Self-hosting in 2026: still worth it?
Vercel and friends keep getting better. Hostinger and friends keep getting cheaper. The choice is no longer obvious.
Five years ago "deploy to a managed platform" was the obvious advice. In 2026 the math has shifted.
The case for managed
Zero ops. Preview URLs out of the box. Edge network without thinking. Time-to-first-deploy in minutes. For solo devs and small teams ship-ship-ship-ing client work, this is still right.
The case for self-hosting
You can rent a serious VPS for the price of a coffee subscription. PM2 + Nginx + Let's Encrypt is a 30-minute setup that runs for years untouched. No vendor surprise pricing. No region lock-in. Logs and database on the same machine, queryable directly.
What we tell clients
If the project will live for 18+ months, has predictable traffic, and the team has any ops capacity at all - self-hosting saves real money. If the project is testing a market or has spiky traffic, managed wins by a wide margin.
We run our own studio site on a VPS. We deploy client work to managed platforms. Both are correct.