A
argbe.tech - news
1min read

Turbopack’s incremental model: faster rebuilds by tracking only what changed

Next.js shared how Turbopack’s incremental computation is designed to keep rebuild work proportional to your edits, not your full app.

The Next.js team published a deep dive on January 20 into how Turbopack’s incremental computation aims to make rebuilds depend on your changes, not the size of the whole application.

  • Turbopack is positioned as the default bundler for Next.js, built to support fast iteration loops and React Fast Refresh at large application scale.
  • The post argues incremental caching can backfire (extra tracking, CPU/memory overhead) if done poorly, but says it’s foundational to Turbopack’s performance goals.
  • Turbopack’s cache is function-level and automatic: it tracks which internal computations depend on which values, so only affected results are recomputed after an edit.
  • It introduces “value cells” (Vc<…>) as a core mechanism for dependency tracking, describing them as spreadsheet-like units that become dependencies when they’re read/awaited.
  • The design draws on prior work and tooling including webpack experience and ideas from Salsa (used by Rust-Analyzer and Ruff), Parcel, Rust’s query system, and Adapton.