TypeScript compiler moves to Go

TypeScript is the checker and tooling layer that sits under a huge chunk of modern JavaScript. When it feels slow, whole teams feel slow — editor startup drags, CI waits pile up, and big monorepos turn every refactor into a coffee break. That’s the backdrop for the real news here: Microsoft didn’t just optimize TypeScript. It moved the compiler and language service onto a new Go-based foundation, and TypeScript 7.0 Beta is now out. (devblogs.microsoft.com) ### What actually changed? For years, TypeScript was a self-hosted toolchain — written in TypeScript, compiled to JavaScript, then run on Node. On April 21, 2026, Microsoft released TypeScript 7.0 Beta and said the codebase had been ported to Go. The key point is “ported,” not “reinvented” — the team says the type-checking logic is(devblogs.microsoft.com)he runtime underneath. (devblogs.microsoft.com) ### Why Go instead of just more tuning? Because the old bottleneck was no longer just a few hot paths. TypeScript has to parse, analyze, and cross-reference a lot of code, and on very large projects the JavaScript implementation hit limits in startup time, memory use, and parallel work. Microsoft’s pitch is that native code plus sh(devblogs.microsoft.com)ch. (devblogs.microsoft.com) ### Are the speed claims real? The official benchmark table is the reason people are paying attention. In Microsoft’s March 11, 2025 post, VS Code dropped from 77.8 seconds to 7.5 seconds, Playwright from 11.1 to 1.1, and TypeORM from 17.5 to 1.3. The company kept using the same rough claim in the beta launch a year later — TypeScript 7.0 (devblogs.microsoft.com) but it does mean this is not a tiny compiler-speed vanity project. (devblogs.microsoft.com) ### What do developers use now? Right now, the beta ships through `@typescript/native-preview`, and the executable is `tsgo`, not `tsc`. There’s also a VS Code extension for the native preview. Microsoft says the package name will eventually become `typescript`, which tells you the endgame pretty clearly — this is meant to become the default toolchain, not a side experiment. (devblogs.microsoft.com) ### Does this break the language? Mostly, the goal is the opposite. The team says 7.0 was ported methodically from the existing implementation, tested against the project’s long-built test suite, and already used on multi-million-line codebases inside and outside Microsoft. So the promise is familiar behavior, just faster and light(devblogs.microsoft.com)s, tooling integrations, and weird build setups. Beta means “ready to try,” not “nothing can surprise you.” (devblogs.microsoft.com) ### What happens to TypeScript 6.0? This part matters more than the benchmark chart. Microsoft opened a January 7, 2026 issue putting 6.0 into maintenance mode and calling it the last scheduled release based on the JavaScript codebase. It also said most open PRs against that old codebase would not be merged, because maintaining two(devblogs.microsoft.com) branch is Go. (github.com) ### Why is this a bigger deal than one compiler? Because TypeScript is not just a compiler. It powers editor intelligence, refactors, navigation, and a lot of the semantic plumbing other tools lean on. If those operations get dramatically cheaper, large codebases stop feeling so heavy. That also fits a broader pattern in developer tooling — performance-critical in(github.com) ones when the ergonomics are worth preserving but latency starts to hurt. (devblogs.microsoft.com) ### Bottom line This is a language-tooling engine swap disguised as a version bump. If Microsoft lands the migration cleanly, TypeScript 7.0 won’t change what developers write nearly as much as it changes how fast their tools can think. (devblogs.microsoft.com)