Every few months a new benchmark chart claims to settle the Vite versus Webpack versus Turbopack debate, and every few months the chart turns out to be measuring the wrong thing. The real story in 2026 is not which bundler wins a synthetic speed test. It is that all three have converged on the same bet, Rust-based compilation, while quietly leaving behind very different lists of what still breaks when you switch.
Key takeaways
- Turbopack became the default, stable bundler in Next.js 16 (shipped October 21, 2025), with Vercel claiming 2 to 5 times faster production builds and up to 10 times faster Fast Refresh, but it still does not support webpack plugins at all.
- Vite 8, declared stable on March 12, 2026, replaced its old split engine (esbuild for dev, Rollup for production) with a single Rust-based bundler called Rolldown, which Vite's own release notes claim is 10 to 30 times faster than Rollup.
- In the State of JS 2025 survey, Webpack still led usage at roughly 86 to 87%, barely ahead of Vite's 84%, but developer sentiment has flipped hard toward Vite.
- Decision rule: if you are not locked into Next.js, default to Vite; move to Turbopack only if you already run Next.js and do not depend on webpack-plugin-only tooling.
The Three Bundlers Are Quietly Converging on Rust
Webpack is written in JavaScript and runs on Node.js, which is part of why it has felt slow for years on large codebases. Vite's original trick was different: it skipped bundling during development entirely and served native ES modules straight to the browser, only falling back to Rollup for production builds. Turbopack took a third path, building a Rust-based incremental compiler from scratch, specifically for Next.js.
That divergence is now closing. Vite 8, announced stable on March 12, 2026, retired the esbuild-plus-Rollup split in favor of a single Rust-based engine called Rolldown, which Vite's own blog post describes as the most significant architectural change since Vite 2. According to that same announcement, Rolldown is 10 to 30 times faster than Rollup and now matches esbuild's speed, with early adopters cited in the post including Linear (build times cut from 46 seconds to 6 seconds) and Beehiiv (a 64% reduction). Webpack remains the outlier, still JavaScript-based, though its plugin ecosystem is precisely why it has not disappeared.
Advertisement
Vendor Benchmarks Have Burned Developers Before
Treat every speed claim in this space with a healthy dose of skepticism, because this industry has a documented history of overselling benchmarks. When Vercel first launched Turbopack in 2022, it published numbers claiming the tool was 10 times faster than Webpack and 5 times faster than Vite, with one chart showing a 700x improvement in hot-module-reload activity over Webpack, according to a detailed writeup by Chariot Solutions.
Vite creator Evan You published his own counter-benchmarks within days. He found the original test had not enabled Vite's SWC-equivalent fast compiler and had skipped server-rendered components, both of which Vercel itself recommends for production Next.js apps. Once configured properly, Turbopack and Vite performed about the same on root-component hot reloads, and Turbopack's real advantage showed up only on deeply nested leaf-component updates, where it was roughly 68% faster, a long way from the headline 5x. The two teams eventually aligned on methodology, but the episode is a useful reminder: a vendor's own number is a starting point, not a verdict.
Is Turbopack Ready to Replace Webpack in Production?
Largely yes for Next.js apps, but not unconditionally. Turbopack reached stability for both development and production builds in Next.js 16 and is now the default bundler with zero configuration needed. Next.js 16.3 went further, experimentally integrating a new Rust-native port of the React Compiler into Turbopack, with Vercel reporting 20 to 50% faster compilation than the older Babel-based version in early tests. Vercel's own Next.js 16 announcement separately claims 2 to 5 times faster production builds and up to 10 times faster Fast Refresh overall.
The catch is its documented gap list: Turbopack does not support webpack plugins at all (only webpack loaders), has no plans to support Yarn Plug'n'Play, and cannot run custom Sass functions because its Rust architecture cannot execute arbitrary JavaScript mid-compile.
Advertisement
There is also a subtler risk buried in Next.js's own Turbopack documentation: Turbopack uses Lightning CSS for styling, which calculates decimal values like line-height to 5 digits of precision instead of webpack's 10. On paper that sounds trivial. In practice it is exactly the kind of silent, pixel-level regression that a design-sensitive team will not catch until a designer flags a site that looks subtly "off" in production, which is the kind of production stability risk that rarely shows up in a benchmark chart.
What the Developer Sentiment Numbers Actually Show
Usage and satisfaction are telling two different stories right now. Both InfoQ's and Heise's independent write-ups of the State of JS 2025 survey, which polled more than 10,000 developers, put Webpack's usage at roughly 86 to 87% against Vite's 84%, essentially a dead heat after years of Webpack dominance. Satisfaction is where the gap opens up. Per Heise's analysis of the same survey, only 14% of Webpack users rated it positively against 37% negative, while 56% of Vite users rated it positively against just 1% negative.
| Bundler | Default engine | 2025 usage share | Plugin compatibility | Known production risk |
|---|---|---|---|---|
| Webpack | JavaScript (Node.js) | ~86-87% | Largest plugin ecosystem in the industry | Slower builds, declining developer satisfaction |
| Vite 8 | Rolldown (Rust) | ~84% | Full Rollup plugin compatibility claimed | Newest engine swap, less production mileage than Webpack |
| Turbopack | Rust, built for Next.js | Not separately tracked outside Next.js | No webpack plugin support, loaders only | CSS precision differences, Yarn PnP unsupported |
That 86-to-84 usage race has been tightening for years. Heise notes the gap was 17 percentage points as recently as 2023 and has narrowed to about 2 points by 2025. If that trend holds, Vite likely overtakes Webpack in raw usage within the next survey cycle or two, even though Webpack's absolute install base is still larger.
✦ Free Newsletter ✦
Never miss a story
Tools, tutorials and AI deep-dives - straight to your inbox, every week.
Who Should Switch, and Who Should Stay on Webpack
This decision depends far more on your dependency tree than on any benchmark. Use this as a starting checklist:
Advertisement
- Choose Vite if you are starting a new project, using React, Vue, Svelte or vanilla JS outside of Next.js, and want the broadest current plugin ecosystem with Rollup compatibility.
- Choose Turbopack if you are already committed to Next.js, do not rely on custom webpack plugins (loaders are fine), and your team can tolerate minor CSS precision differences during migration.
- Stay on Webpack if your build depends on specific webpack plugins with no Rollup or Turbopack equivalent, you use Yarn PnP, or you have a large legacy app where the migration risk outweighs the build-time savings.
- Skip switching entirely if your current build times are not actually a bottleneck for your team; a faster bundler will not fix a slow CI pipeline, bloated dependency graph, or an untested deploy process.
Teams coming from Node, Deno or Bun runtime decisions will recognize the pattern: the tool with the flashiest benchmark is rarely the one that saves the most engineering time once migration costs are counted.
The Honest Case for Sticking With Webpack
The strongest argument against switching is not nostalgia, it is risk management. Webpack has a 12-year track record, an enormous plugin catalog covering nearly every build requirement imaginable, and predictable behavior that thousands of production apps depend on today. Migrating a large, plugin-heavy codebase to Turbopack or Vite is real engineering work with real regression risk, and the time saved on individual builds has to be weighed against the hours spent debugging a migration.
It is also worth noting that both Vite's Rolldown engine and Turbopack's production-build stability are genuinely new. Rolldown only reached its own 1.0 milestone in May 2026, and Turbopack's production-build path has had barely a year of default usage at the time of writing. Webpack's maturity is a real feature, not just inertia, especially for teams that cannot afford a broken release pipeline. If your builds are already fast enough and your team is small, the safest move may be no move at all.
Common Questions
Can I use Turbopack outside of Next.js? Not practically today. Turbopack is built into and tuned for Next.js specifically, and Vercel has not shipped a standalone, framework-agnostic release, so Vite remains the default choice for non-Next.js projects.
Do I need to rewrite my webpack config to switch to Vite? Usually yes, at least partially. Vite uses its own plugin API and configuration format, though many common webpack loaders have direct Vite plugin equivalents for CSS, images, and TypeScript.
Advertisement
Will Webpack be deprecated? Not in the foreseeable future. Webpack remains the default in many non-Next.js setups and continues to receive maintenance, even as its market share slowly erodes.
The practical takeaway: do not pick a bundler based on the biggest multiplier in a release-day blog post. Check your dependency tree for webpack-plugin-only packages first, confirm whether your framework already forces a choice (Next.js effectively picks Turbopack for you now), and only then weigh the Vite-versus-Webpack decision on your own build times, not someone else's benchmark.
Sources
Advertisement
