Hermes v1 Becomes the Default

React Native 0.84 marks an important step forward with Hermes v1 enabled by default. The new engine version brings measurable performance improvements and reduced memory usage, while aligning more closely with modern JavaScript standards.

Kewin and Danny discussed how Hermes v1 builds on earlier experimental efforts such as Static Hermes. With broader ES feature support, some Babel transformations become unnecessary, simplifying the toolchain. For teams maintaining larger apps, this reduces configuration overhead and moves React Native closer to parity with modern web runtimes.

The takeaway is practical: better defaults, fewer workarounds, and a more capable JavaScript engine out of the box.

Faster Builds With iOS Precompiled Binaries

One of the most developer-visible improvements in 0.84 is precompiled React Native binaries on iOS.

Previously, large portions of React Native were rebuilt during every iOS compilation, even when the source code hadn’t changed. With prebuilt binaries, much of that repeated work disappears. Build times improve, iteration becomes smoother, and the development loop tightens.

This change also reflects broader ecosystem maturity. As core internals stabilize, more components can safely ship in compiled form, reducing friction without sacrificing flexibility.

WebAssembly Support in Hermes

WebAssembly support landing in Hermes opens new integration paths for performance-sensitive code.

With WASM available, developers can compile Rust, C, or C++ to WebAssembly and execute it within React Native. This lowers the barrier for reusing existing native logic and unlocks advanced use cases such as on-device AI inference, heavy numerical computation, or graphics processing.

Instead of rewriting performance-critical code in platform-specific languages, teams can target WebAssembly and run it directly inside the JavaScript runtime. It’s a significant step toward making React Native a more flexible execution environment.

Tooling Evolution and Ecosystem Direction

Beyond core changes, the conversation expanded into the surrounding ecosystem.

ESLint v9 migrations, alternative tooling stacks like Oxc and Bun, and growing interest in Rust-based infrastructure signal a broader shift. Performance, stability, and developer ergonomics are driving decisions at every layer, from bundlers to linters to runtime engines.

Danny also highlighted ongoing experimentation around developer tooling, including Storybook integrations and ideas for AI-assisted UI iteration. The direction is clear: tighter feedback loops and better isolation of UI components will play an increasingly important role in modern React Native workflows.

React Native AI and On-Device Models

The discussion touched on React Native AI efforts, including support for on-device models via MLC and compatibility with common AI SDK patterns.

Running inference locally introduces new constraints (memory, thermal limits, battery consumption), but also enables privacy-preserving and offline experiences. With Hermes gaining WebAssembly support and React Native libraries embracing local inference, the groundwork is being laid for more capable mobile AI features.

The ecosystem is not only adopting AI APIs; it’s exploring deeper runtime-level integrations.

Watch the full stream recording to explore the technical details behind React Native 0.84 and the evolving tooling landscape.

‍