Summary
Vercel was easy for one framework. With new services coming, they wanted a platform that wouldn't decide their stack for them.
The challenge
On Vercel they paid per project and could add a database and auth, but it wasn't flexible enough to run a different stack. At the proof-of-concept stage that didn't matter: Next.js, with the database on Supabase, one stack.
As the product grew, new services were coming, so they started looking for something else.
What they tried
Cloudflare meant a hard Next.js setup and a whole evening spent deploying. The other options looked like a hyperscaler with a price list you need a PhD to read, or a dedicated server, with its setup, its upkeep and a single point of failure.
Why Miget
On Miget they could click everything together and deploy it themselves, quickly. Load balancers, TLS and autoscaling stopped being their problem: they push a change, it loads, and they don't need to know exactly how it's hosted.
They didn't want to fit the product to the infrastructure, they wanted to stay infrastructure-agnostic. With a Docker image, they could guarantee that on Miget.