The bundler swap nobody has to opt into
Vite 8 shipped on March 12th, and the headline change is architectural rather than feature-based: Rolldown, the Rust-based bundler from Evan You's VoidZero team, is now the default under the hood, replacing both esbuild (which handled dev-mode transforms) and Rollup (which handled production bundling) in one move.
If you upgrade to Vite 8, Rolldown is simply on. No config flag, no opt-in flag, no migration guide to follow line by line. That's a deliberate choice — the Vite team spent the better part of a year making sure the Rollup plugin API surface stayed compatible enough that most existing plugin ecosystems, including framework-specific ones, keep working without modification.
What actually changed under the hood
Before Vite 8, your dev server and your production build technically ran through two different bundlers with two different sets of quirks — fast but slightly different in dev, slower but "real" in production. Rolldown collapses that gap. One bundler, two modes, and — per the docs — largely the same output semantics between them, though not a strict 1:1 guarantee yet.
Built into the pipeline is Oxc, the Rust-based transformer, handling TypeScript, JSX, and modern syntax compilation without needing a separate plugin. CommonJS interop, previously requiring @rollup/plugin-commonjs, is now handled natively by Rolldown following esbuild's interop conventions.
Running it against a real app
I tested the upgrade against a mid-sized React app — a few hundred components, heavy use of dynamic imports, a fairly typical plugin configuration for a production SaaS frontend. Production build times dropped noticeably, in line with the broader benchmark trend the ecosystem has been reporting since Rolldown's 1.0 release candidate in January. Dev server cold start felt similar, since esbuild was already fast there; the bigger win showed up in full production builds, where Rollup had always been the slower half of the equation.
Nothing broke on first run. The one snag was a custom plugin relying on a Rollup-specific hook that hadn't been fully ported — a good reminder to check your plugin list against Rolldown's compatibility notes before rolling this out to a whole team, even though the common path is smooth.
Should you upgrade now?
For most projects, yes, with the usual caveat: run it on a branch first, and pay attention to any plugins doing something unusual with the build pipeline rather than just transforming files. For anything shipping today, the risk is low and the upside — meaningfully faster production builds with functionally no config changes — is hard to pass up.