Next.js 16.2 dropped in March with a bold claim: next dev is 400–900% faster. Turbopack is now the default bundler. Server Components get hot reload.
But the number isn't the interesting part.
The boring answer: yes, it's faster
The Vercel benchmarks are real. On projects with 40+ components, cold starts that used to take 7–9s now land around 1–2s. Not magic — Turbopack eliminates Webpack's overhead on module graph analysis.
| Project size | 15.x cold start | 16.2 cold start | Gain |
|---|---|---|---|
| Small (~15 components) | 3.5s | 0.9s | ~4x |
| Medium (~40 components) | 7s | 1.5s | ~5x |
| Large (monorepo) | 12s+ | 2.5s | ~5x |
What actually changes day-to-day
HMR stopped being a "step". Editing a prop in a deeply nested component used to produce a noticeable lag — 200–400ms before the browser caught up. On 16.2 you edit and the screen already changed. The feedback loop gets short enough that you start iterating differently: less "I'll batch these edits to avoid a reload", more "changed it, saw it, adjusted."
Type errors no longer block the server. Turbopack runs type-checking asynchronously. If your types are wrong, you still get the error — but the dev server doesn't hang waiting for the compiler before serving the page. Small change, big difference mid-refactor.
Server Components finally hot-reload properly. "Server Fast Refresh" is the least-talked-about feature in 16.2. Not perfect — moving between RSC and Client still occasionally needs a full reload — but solid enough to stop being annoying.
What they don't mention (the trade-offs)
If you have custom Webpack plugins — analyzers, module federation, exotic loaders — check Turbopack parity before upgrading. Most common ones are covered; niche setups may need migration work.
Also: config that worked in Webpack can behave differently in Turbopack. The most common symptom is something working in dev and breaking in next build. Always run a full production build before deploying after an upgrade.
Should you upgrade?
- On 15.x → yes, today. Budget 30 min for a small app.
- On 14.x → go in stages: 14 → 15.x → 16.2. Don't skip.
- On 13.x with
pages/→ this is a larger migration. App Router + Turbopack at the same time deserves a dedicated sprint.
The "400% faster" number is real. But the actual unlock is subtler: when the feedback loop shrinks, you try more things. That's what Vercel doesn't put in the changelog.
Migrating to 16.2 — what went smoothly and what didn't for you? 👇