Angular v22 graduates Signal Forms and async reactivity, closing the post-Zone.js pivot
Angular v22 shipped stable on 2026-06-03, graduating Signal Forms, the async reactivity APIs (resource, httpResource), and Angular Aria out of experimental. The version where the post-Zone.js pivot becomes the whole stack, not a partial recommendation.
Angular shipped v22 as stable on 2026-06-03, graduating Signal Forms,
the async reactivity APIs (resource, httpResource), and Angular Aria
out of experimental. The release is the version where the multi-year
post-Zone.js pivot stops being a partial recommendation and starts
being the whole stack. The pitch for new apps is no longer “drop Zone
and see what happens.” It is a typed reactive surface designed to live
together.
What shipped
Signal Forms is the headline. It rebuilds Angular’s forms primitive on signals, with typed reactive forms, custom validation, dynamic bindings, and accessibility integration through Angular Aria. Forms have been one of Angular’s most distinctive surfaces since the framework’s relaunch a decade ago, and they are also the surface most teams have wired through RxJS observables. Signal Forms is the explicit replacement, designed to keep the typed-binding ergonomics while moving the underlying graph onto the same signal model the rest of the framework now uses.
The async reactivity APIs graduate alongside Signal Forms. resource
and httpResource let server-data fetches participate in the same
signal graph as local state, with a declarative HTTP call that re-fires
when a dependent signal changes. Together with Signal Forms, the two
close the loop on the post-Zone.js story Angular has been building since
v17, when standalone components first shipped, through v21’s
zoneless-by-default change detection.
Angular Aria moves to stable in the same release. It covers accessible UI patterns (menu, listbox, dialog, combobox) with Signal Forms-aware bindings and test harnesses for keyboard, mouse, screen-reader, and custom input paths.
Template syntax picked up smaller but useful refinements: multi-case
matching to consolidate switch arms, exhaustive type checks on union
defaults when the never keyword is used in the default case, and
inline arrow functions inside template expressions. Following
deprecations begun in v20, the structural directives ngIf, ngFor,
and ngSwitch are now officially deprecated in favor of the
control-flow syntax (@if / @for / @switch) that shipped earlier.
The control-flow blocks have been the recommended path since v17. v22
makes the legacy directives an explicit migration target.
AI tooling expanded with updated MCP dev server tools, Agent Skills for coding assistants, Google AI Studio and Gemini Canvas integration, and experimental WebMCP support. The v22 release video airs 2026-06-05 on the Angular YouTube channel.
Where this lands in the market
Angular’s repositioning over the v17-to-v22 stretch has been one of the more decisive framework rewrites of the last decade. The framework that defined NgModules, Zone.js, and RxJS-everywhere now has a primary recommendation that uses none of those by default. v22 is the version where the recommendation feels complete, not partial.
The Signal Forms graduation matters most for teams running production apps on reactive forms with RxJS observables. The upgrade path no longer asks them to stay on the old form model while the rest of the app moves to signals. That mismatch has been the largest reason established Angular shops have hesitated to commit to the new direction, and v22 is the first release where committing all the way through is the path of least resistance rather than the leading edge.
The comparison frame that matters is React, not Vue or Svelte. React’s
own signal-adjacent direction lives in React Compiler and refinements
to useTransition, with no signal primitive in core. Angular and Solid
sit on the same side of the signals debate, with concrete signals in
core, while React has chosen to optimize the existing useState model
through compilation. v22 hardens Angular’s bet on that fork.
What’s worth watching
-
Signal Forms adoption in the enterprise long tail. Angular’s enterprise install base is the largest reason the framework still ships a v-major release video in 2026. Whether Signal Forms migrations land cleanly in legacy reactive-forms codebases is the real test of whether v22 is a viable upgrade target or a starting-fresh-only release. The migration codemod story will be the first signal.
-
ngIf/ngFor/ngSwitchremoval timing. The trio are deprecated in v22 but not removed. The Angular team’s typical deprecation cycle is two major releases. If v23 ships removal alongside a fully baked migration codemod, the structural-directive era ends cleanly. If removal slips, it suggests the migration story has gaps the team is not ready to expose. -
WebMCP and Agent Skills traction beyond Google’s stack. v22’s AI tooling pitch leans heavily on MCP and Google’s agent-skill model. If the surface picks up coding-assistant integrations beyond Gemini and Claude over the next two quarters, it becomes a defensible framework-side moat. If it stays Gemini-flavored, it stays a Google ecosystem play.
The follow-up beat lands around 2026-09-02, after one full quarter of v22 migration data is visible.
- Angular blog: Announcing Angular v22 blog.angular.dev
- Angular v22 Release event angular.dev
- AlternativeTo: Angular v22 launches with Signal Forms, AI Tooling, and Async Reactivity APIs alternativeto.net