Back to Native: How Coding Agents Are Reversing the Cross-Platform Trend
Shopify is returning to native Swift and Kotlin after six years of React Native — citing coding agents as the reason. Here's why this matters beyond a single case study.
In 2020, React Native was a rational choice for Shopify: one team, one codebase, halved development costs across two platforms. In September 2026, Shopify reversed course and returned to native Swift and Kotlin — with a reasoning that should make everyone pay attention: the main advantage of cross-platform frameworks has disappeared.
The 2020 math doesn’t work in 2026
Cross-platform frameworks live off a simple equation: write once, ship twice. Duplicated feature implementation for iOS and Android was the main cost driver of mobile development — React Native, Flutter and friends promised to cut it in half.
This is exactly where coding agents strike. An agent implements a feature once for iOS, translates it to Android, adapts testing and review processes across platforms — in a fraction of the time human teams need. Shopify demonstrated how fast this has become: the entire Shop app was rebuilt natively in just twelve weeks with AI assistance and is already live in the app stores. What used to be a six-month, two-team project is now a one-quarter, one-team project.
The core thesis: agents make platform costs predictable
The real story is not “React Native is dead”. The real story is: the cost advantage of “write once, run twice” is measurably shrinking, because translating between platforms — historically the expensive part — has become largely automatable.
When features can be translated, tested and reviewed across platforms automatically, the framework is no longer destiny. Teams get to choose based on other criteria: performance, platform integration, tooling, long-term maintainability. Native often wins that trade-off — not ideologically, but economically.
What this means for studios
For agencies like us, there is a wrong and a right conclusion.
The wrong one: “All the more reason for React Native — agent support exists everywhere now.” The right one: “The framework is no longer a one-way street. We choose per project — and can even change direction mid-project if conditions change.”
Concretely, for client work: mobile decisions no longer last forever. An MVP can start cross-platform for speed; once it proves itself, a native rebuild with agent assistance is no longer a nightmare but a calculable quarter.
Outlook: the ecosystems are converging
Interestingly, both camps are moving toward each other at the same time: Swift 6.4 (mid-September 2026) makes Swift Build the default and brings official Android support via NDK 30 — the native iOS language running on Android. And Kotlin 2.4.20 ships native Swift Export for Kotlin Multiplatform — the JVM language exporting cleanly to Swift. Both ecosystems are preparing for developers working in both worlds — often with an agent as the translator in between.
If you’re starting a mobile project today, treat the framework question more casually than you would have in 2024: decide per project, let agents do the translation work, and treat technology for what it is again — a tool, not a creed.