Shopify spent six years telling the industry that React Native was the future of mobile at their scale. In January 2025 they wrote a whole post reaffirming it. Eight months later they published "Native is now the future of mobile at Shopify" and quietly buried the framework that shipped their flagship apps to millions of merchants.
If you skim the HN thread (1,000+ points, 700+ comments as of this morning) you'll see the usual pile-on: "React Native was always a mistake," "told you so," "cross-platform is a lie." That's the wrong takeaway. Read the actual post and the real story is sharper — and scarier if you're a platform-team lead who just bet your roadmap on Flutter or RN.
The numbers, because they matter
Shopify rebuilt their Shop app (~2M downloads/week) from a blank proof-of-concept to a published App Store release in 12 weeks, with a team of six engineers. The results, per their own engineering blog:
- Session crash-free rate jumped to 99.95%+, a 10x reduction in crashes vs. the 99.5% baseline they'd been living with on React Native
- The Android release build shrank by 109 MB
- Android build time dropped ~75%
The Shopify app — the big one, 300+ screens, home/lock-screen widgets, an Apple Watch app, Siri Shortcuts — is mid-migration right now. React Native gets end-of-life support through the end of 2026, then it's gone from Shopify entirely.
Here's the part everyone's missing
Nobody at Shopify is claiming React Native got worse or that cross-platform frameworks are fundamentally broken. Their own words:
"coding models have gotten dramatically better... building the same feature in Swift and Kotlin no longer carries the cost it used to"
That's the whole thesis. In 2020, "write once, run everywhere" won because a human engineer writing the same feature twice, in two languages, on two platforms, was expensive. In 2026, that premise is dead — not because RN got slower, but because the marginal cost of writing native code twice collapsed.
Shopify built a system called Helix specifically for this: migrations get broken into checkpoints, each one requiring passing tests, visual diff verification, adversarial code review, and human sign-off before the agent moves to the next chunk. They also decoupled business logic into a CLI layer so agents can iterate at millisecond speed without booting a simulator every time. That's not "vibe coding a migration" — that's an actual engineering system built around the assumption that agents do the grunt work and humans gate the checkpoints.
This is the tell. The interesting shift in 2026 isn't "AI writes code now" — we knew that. It's that AI writing code changed which architectural tradeoffs are even worth making. Cross-platform abstraction layers exist to amortize human labor across platforms. When the labor cost approaches zero, the abstraction stops paying for itself, and you're left holding its downsides for free: an extra runtime layer, bridge overhead, platform APIs that lag by a release cycle, and — this is the one nobody likes to say out loud — a debugging experience that's always one layer removed from what's actually crashing.
What this means if you're not Shopify
Before you rage-quit your RN or Flutter codebase, know what Shopify actually had that most teams don't:
- Six years of production telemetry on exactly where React Native cost them — not vibes, actual crash logs and build metrics
- A dedicated platform team that built custom tooling (Helix, the CLI decoupling layer) instead of just pointing Cursor at the repo and hoping
- The scale to justify it — 2M+ weekly downloads on one app alone means a 10x crash reduction is worth a 12-week rebuild in a way it isn't for your 50k-MAU app
If your team is three people maintaining a cross-platform app because you don't have the headcount to duplicate feature work, agents lower that bar too — but "lower" isn't "zero," and you don't have Shopify's Helix pipeline sitting around to catch regressions during the rewrite.
The actual lesson: go re-run your own cost-benefit math, because one of your inputs just changed by an order of magnitude. Cross-platform vs. native was never a religious argument, it was an economics argument, and the economics just moved. Shopify didn't discover that native is "better." They discovered that the reason they picked React Native in 2020 stopped being true.
If you're maintaining react-native-skia or FlashList and wondering what happens next: Shopify is sponsoring Skia through end-2026 with William Candillon forking it, looking for a new steward for FlashList, and archiving Restyle outright. That's the unglamorous part of this story — someone still has to own the libraries the rest of the ecosystem depends on, and "we don't need this anymore" from your biggest corporate sponsor is a real problem for maintainers, agents or no agents.
Watch what the rest of the industry does over the next two quarters. If Airbnb, Coinbase, or another RN-at-scale shop makes the same call citing the same reasoning, this stops being a Shopify anecdote and becomes the first real data point that AI coding agents are quietly rewriting which architecture patterns are economically rational — not just which ones are fashionable.
Top comments (0)