Meta just merged something significant into the React repository: a complete Rust rewrite of the React Compiler. Not an experiment in a separate branch \u2014 merged into the main React monorepo, shipping in Next.js 16.3, and already delivering 20\u201350% faster builds in real applications.<\/p>\n
If you’ve been watching the frontend toolchain shift toward Rust over the past two years \u2014 SWC replacing Babel, Turbopack replacing webpack, Biome replacing ESLint \u2014 this is the next piece of that picture snapping into place.<\/p>\n
Before getting into the Rust rewrite, a quick refresher on what the React Compiler is.<\/p>\n
React Compiler (formerly called React Forget) automatically memoizes your components and hooks. In plain English: it makes React smarter about when to re-render components, without you having to manually sprinkle The compiler analyzes your code and inserts the memoization automatically at build time. Your code is cleaner. Runtime performance is the same or better. You stop writing The original React Compiler was written in TypeScript. This made sense for developer familiarity and iteration speed \u2014 the React team could ship and iterate quickly.<\/p>\n But TypeScript compilers run on Node.js, and Node.js has overhead that Rust doesn’t. As the React Compiler integrated deeper into the build pipeline \u2014 running on every file, every build, in watch mode \u2014 that overhead became a real cost.<\/p>\n The Rust rewrite has a specific technical advantage beyond raw speed: it can be linked directly into Turbopack (Vercel’s Rust-based bundler) without going through a serialization boundary. Previously, running the compiler as a Babel plugin meant data had to cross from JavaScript into the plugin and back \u2014 adding overhead on every file transformation. The Rust version eliminates this entirely.<\/p>\n Performance improvements from the Rust port:<\/p>\n All 1,725 test fixtures pass. Intermediate states match the TypeScript version almost byte for byte. The migration isn’t just fast \u2014 it’s correct.<\/p>\n What changed:<\/strong> The internals. The compiler is now rebuilt with arena allocation and index-based data structures \u2014 Rust memory patterns that avoid the garbage collection overhead of JavaScript runtimes.<\/p>\n What didn’t change:<\/strong> The public API. The configuration is identical. Upgrading is intended to be a drop-in swap.<\/p>\n If your codebase has components that aren’t compiler-compatible (those that violate React’s rules), you can enable the compiler selectively:<\/p>\n This isn’t an isolated event. It’s part of a deliberate shift across the entire JavaScript toolchain toward Rust:<\/p>\n The pattern is consistent: JavaScript tooling works, then gets rewritten in Rust when it becomes a bottleneck. The React Compiler joining this list is a signal that build performance is now a first-class concern, not an afterthought.<\/p>\n The InfoQ report noted a fair criticism from the developer community: this rewrite leaned heavily on LLMs for the mechanical porting work. Humans handled architecture and review, but the bulk of the code was AI-generated.<\/p>\n One Hacker News commenter raised the obvious question: if no human deeply understands the implementation, what happens when it breaks in a subtle way?<\/p>\n This is a genuine concern. Rust that compiles doesn’t mean Rust that’s well-written \u2014 a model can satisfy the borrow checker with The 1,725 passing tests are reassuring. Production use over the next 6 months will tell the real story.<\/p>\n Practically \u2014 if you’re building React frontends with Laravel and Vite:<\/p>\n The build time improvements are real and compound over a large codebase. A 40% faster build means your developers spend less time waiting and more time writing code. For teams doing continuous integration, it also means cheaper CI bills.<\/p>\nuseMemo<\/code> and useCallback<\/code> everywhere.<\/p>\n\/\/ Before React Compiler \u2014 you manually memoize\r\nfunction ExpensiveComponent({ data, filter }) {\r\n const filtered = useMemo(\r\n () => data.filter(item => item.category === filter),\r\n [data, filter]\r\n );\r\n\r\n const handleClick = useCallback(() => {\r\n console.log(filtered);\r\n }, [filtered]);\r\n\r\n return <List items={filtered} onSelect={handleClick} \/>;\r\n}\r\n\r\n\/\/ With React Compiler \u2014 the compiler handles this automatically\r\nfunction ExpensiveComponent({ data, filter }) {\r\n const filtered = data.filter(item => item.category === filter);\r\n\r\n const handleClick = () => {\r\n console.log(filtered);\r\n };\r\n\r\n return <List items={filtered} onSelect={handleClick} \/>;\r\n}<\/code><\/pre>\nuseMemo<\/code> by instinct.<\/p>\nWhy Rewrite It in Rust?<\/h2>\n
The Numbers<\/h2>\n
\n\n
\n \nScenario<\/th>\n Improvement<\/th>\n<\/tr>\n<\/thead>\n \n As a Babel plugin drop-in<\/td>\n ~3x faster than TypeScript version<\/td>\n<\/tr>\n \n Isolated transformation logic<\/td>\n Up to 10x faster<\/td>\n<\/tr>\n \n Integrated with Turbopack (v0 app)<\/td>\n >40% faster compilation<\/td>\n<\/tr>\n \n Next.js 16.3 test apps<\/td>\n 20\u201350% faster route compilation<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n What Changed, What Didn’t<\/h2>\n
\/\/ vite.config.ts \u2014 same config, faster compiler\r\nimport { defineConfig } from 'vite';\r\nimport react from '@vitejs\/plugin-react';\r\n\r\nexport default defineConfig({\r\n plugins: [\r\n react({\r\n babel: {\r\n plugins: [\r\n ['babel-plugin-react-compiler', {\r\n target: '19', \/\/ same as before\r\n }],\r\n ],\r\n },\r\n }),\r\n ],\r\n});<\/code><\/pre>\nUsing the React Compiler Today<\/h2>\n
With Next.js (Experimental in 16.3)<\/h3>\n
\/\/ next.config.ts\r\nimport type { NextConfig } from 'next';\r\n\r\nconst config: NextConfig = {\r\n experimental: {\r\n reactCompiler: true,\r\n },\r\n};\r\n\r\nexport default config;<\/code><\/pre>\nWith Vite + React<\/h3>\n
npm install babel-plugin-react-compiler<\/code><\/pre>\n\/\/ vite.config.ts\r\nimport { defineConfig } from 'vite';\r\nimport react from '@vitejs\/plugin-react';\r\n\r\nexport default defineConfig({\r\n plugins: [\r\n react({\r\n babel: {\r\n plugins: ['babel-plugin-react-compiler'],\r\n },\r\n }),\r\n ],\r\n});<\/code><\/pre>\nIncremental Adoption<\/h3>\n
\/\/ Opt out a specific file\r\n\/\/ @ts-nocheck\r\n'use no memo'; \/\/ Tells React Compiler to skip this file\r\n\r\n\/\/ Or configure globally with an allowlist\r\n\/\/ next.config.ts\r\nconst config: NextConfig = {\r\n experimental: {\r\n reactCompiler: {\r\n compilationMode: 'annotation', \/\/ Only compile annotated components\r\n },\r\n },\r\n};\r\n\r\n\/\/ Then in components you want compiled:\r\n\/\/ 'use memo'; \/\/ Opt this component in<\/code><\/pre>\nThe Bigger Picture \u2014 Rust Is Eating the Frontend Toolchain<\/h2>\n
\n\n
\n \nTool<\/th>\n JavaScript Version<\/th>\n Rust Replacement<\/th>\n<\/tr>\n<\/thead>\n \n Transpiler<\/td>\n Babel<\/td>\n SWC \u2705 (widely deployed)<\/td>\n<\/tr>\n \n Bundler<\/td>\n Webpack<\/td>\n Turbopack \u2705 (Next.js), Rolldown (Vite)<\/td>\n<\/tr>\n \n Linter\/Formatter<\/td>\n ESLint + Prettier<\/td>\n Biome \u2705 (growing fast)<\/td>\n<\/tr>\n \n React Compiler<\/td>\n TypeScript<\/td>\n Rust \u2705 (just merged)<\/td>\n<\/tr>\n \n Type Checker<\/td>\n tsc<\/td>\n ts-go (Microsoft, in progress)<\/td>\n<\/tr>\n \n Package Manager<\/td>\n npm<\/td>\n bun \u2705, cargo (for WASM targets)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n The Legitimate Concern<\/h2>\n
RefCell<\/code> and push failures to runtime. Whether this specific codebase has those problems is unknown until it gets production load and edge case exposure.<\/p>\nWhat This Means for Laravel + React Developers<\/h2>\n
\n
useMemo<\/code> and useCallback<\/code> by instinct \u2014 the compiler handles this better than manual annotation.<\/li>\n<\/ul>\n