- Lovable is replacing Vite's development server with OJ, a Rust tool, in its cloud previews. Lovable says this cut median sandbox start time from 14.5 to 3 seconds and memory about 6.5x.
- Vite's creator called Lovable's first speed test misleading. In bundled mode, cold start was 1,315 ms for OJ versus 1,528 ms for Vite, though memory was 115 MB versus 1,751 MB.
- Vite's creator expects AI to make company-specific forks common, and a fix in Vite may not reach them. Leaders should list internal forks and confirm someone tracks upstream security patches.
A Dev Server Rebuilt for One Company's Scale
Lovable, a platform for building applications with AI, has started swapping Vite out of its cloud preview environments in favor of OJ — a build tool written in Rust to handle roughly a million short-lived development sandboxes the company spins up daily. OJ is not a full rewrite of Vite. Only the development-server layer changed hands: file watching, the module graph, compiler coordination, hot updates, and WebSocket handling now run through OJ's Rust binary, while bundling and parsing still run on Rolldown and Oxc, the same engines already inside Vite 8. "It is not an entire rewrite of Vite," Vite creator Evan You said on X.
The Speed Gap Narrows. The Memory Gap Doesn't.
Lovable's first published benchmark measured OJ running in bundle mode against Vite running in its default, unbundled mode, on a test app built with 10,000 React components; OJ reached a cold start in roughly 1.2 seconds versus 4.9 seconds for Vite. Vite's creator pushed back, calling the test "somewhat misleading" since Vite also ships a bundled development mode that wasn't used in that comparison. Once both tools were measured in bundled mode, the cold-start gap nearly closed: 1,315 milliseconds for OJ versus 1,528 milliseconds for Vite. Separately, Lovable acknowledged that one of its real-world test cases ran Vite with the plugin vite-plugin-checker installed, which spins up a TypeScript-checking worker process; OJ has no equivalent yet and simply skips that step, a gap that undercuts the largest speed claim while leaving the memory comparison untouched. OJ creator Raphael Amorim said the tool has been tested against production-scale codebases, including Excalidraw and Twenty, whose frontend alone contains roughly 15,000 modules, and that "both run on OJ unchanged."
AI Lowers the Cost of Forking Shared Infrastructure
OJ's arrival fits a pattern already underway across JavaScript tooling: pnpm rewrote its own package manager in Rust, and other widely used projects — SWC, Biome, Oxc, Rolldown, and portions of Turbopack — have each swapped out JavaScript components for Rust in pursuit of speed. Evan You frames OJ's distinguishing trait differently: rather than contributing improvements back to a shared project, Lovable rebuilt a single layer of that general-purpose tool around the specific needs of its own sandbox infrastructure. You said he expects AI tooling to make this kind of project far more common, citing TanStack's Redact — a smaller, synchronous replacement for React's runtime that keeps React's own APIs intact — as another example of what he called a "tailored projection." The cost is fragmentation: a fix shipped inside Vite's own development server does not automatically reach a Vite-compatible fork, and the meaning of "compatible" can vary from one reimplementation to the next. As AI lowers the cost of translating an established project into a narrower, company-specific version, more of the engineering work behind widely used developer infrastructure could move into separately maintained forks even as the original project's interfaces remain the standard.
What Budget Holders Should Ask
The operative question for organizations that depend on AI-assisted development platforms, or that maintain their own build and preview infrastructure, is not whether a Rust rewrite runs faster. It is whether any team has quietly forked shared tooling to solve a scaling problem, and who owns the job of pulling upstream security fixes into that fork once it diverges. Budget holders overseeing engineering and security spend should ask their teams to list any internally maintained forks or reimplementations of open-source build tools or frameworks, and confirm there is a defined process for tracking upstream patches once a compatibility layer exists between the fork and the original project.
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: Socket.





