Why Developers Are Leaving Vercel — and Where They're Going

Something shifted in the developer community over the past year. The conversation around Vercel changed from "this is amazing" to "did you see my bill?" What was once the default deployment platform for Next.js and modern frontend apps is now the subject of Reddit threads, Hacker News discussions, and blog posts about surprise invoices and pricing traps. This is not about Vercel being a bad product. It is an excellent deployment platform with a superb developer experience. The issue is the gap between the experience of using Vercel and the experience of paying for it -- a gap that widens significantly the moment your app gets real traffic. … - **Build minutes:** 6,000 minutes included. Overages at $0.01 per minute. A Next.js app with a large page count or heavy static generation can consume 5-10 minutes per build. Push 20 times a day across a team and you are burning through those minutes fast. Some teams report hitting the ceiling within the first two weeks of the month. - **Serverless function invocations:** 1 million included on Pro. Every SSR page render, every API route call, every ISR revalidation counts as an invocation. A moderately trafficked e-commerce site can exhaust this in days. - **Bandwidth:** 1TB included. Overages at $0.15 per GB. If you serve images, videos, or large JSON payloads, 1TB is not generous. One viral moment and your bandwidth bill spikes. … The result is predictable: developers share stories of $400 surprise bills, $1,200 monthly invoices for what they expected to be a $20/month service, and the realization that the "serverless" model means you are paying per-request for things that a persistent server handles for a flat rate. The most frustrating aspect is not the cost itself -- it is the unpredictability. With usage-based pricing, a traffic spike (which should be a good thing) becomes a financial risk. You cannot budget accurately because you cannot predict your bill.