Next.js 15.1 is unusable outside of Vercel

Original Article ↗ Hacker News Discussion ↗

Frustration with Next.js 15.1 and Vercel-Centric Design

  • Many see 15.1’s streaming metadata change (sending meta tags late, via an “htmlLimitedBots” user-agent list) as a breaking SEO regression that mainly benefits a tiny subset of users.
  • Criticism that this should have been opt‑in, or at least a clearly named hard opt‑out, not hidden behind a bot-focused option.
  • Combined with other regressions (e.g., broken middleware prefetch detection, libraries breaking after 15.1.8), people describe “Next.js fatigue” and a sense that the framework is now brittle and over‑magical.

Developer Experience and Performance Issues

  • Multiple reports of 10+ second rebuilds per route in dev, with next dev --turbo not helping; compared unfavorably to Rust/C++ compile times.
  • Complaints that Next runs too much user code at build time, has confusing “DynamicError”-style failures, and that docs and example Dockerfiles can be misleading or hard to use.
  • The <Image> component is mentioned as a frequent source of opaque performance problems, including a severe FPS drop in one WebGL case.

Hosting, Lock-In, and Scale

  • One camp: “Just Dockerize and deploy, it’s fine;” they argue confusion comes from front-end devs lacking infra skills.
  • Another camp: self-hosting is fragile, especially with serverless features; vendor lock-in to Vercel is seen as deliberate strategy.
  • Large-scale users report that major upgrades can take months on big monorepos, and that cracks appear when pushing Next hard.
  • A counterexample describes successful large-scale RSC + OpenNext + AWS Lambda + custom CDN/Redis setup, with big cost savings and good SEO metrics.

RSCs and Framework Direction

  • Some stay on Next solely for React Server Components, which they find powerful.
  • Others find RSC boundaries and the App Router/Server Actions “magic” confusing, overcomplicated, and ill-suited for SPAs; many argue most web apps don’t need SSR/streaming.

Alternatives and Recommendations

  • Frequently suggested replacements: React + Vite (+ React Router or TanStack Router/Query/Form, SWR), Astro (especially for content-heavy or static sites), React Router 7 “framework mode,” TanStack Start (though lacking RSC), and more “boring” setups (SPA + separate backend).
  • Several teams report switching from Next to Astro or Vite and feeling immediate relief in speed, simplicity, and deploy flexibility.

Ecosystem Power and Trust Concerns

  • Complaints about Vercel’s aggressive brand management (reaching out over critical posts), growth-hacking style outreach, and strong influence over React docs and defaults.
  • Some frame Vercel’s strategy as building a vertically integrated, React-centric hosting lock‑in, prompting calls to be cautious with any Vercel-affiliated stack.