Recent Shopify’s announcement caught me by surprise: they’re going to rewrite all their apps from React Native back to native mobile stacks, less than two years after they announced that their five-years-long migration to React Native had been a tremendous success.
The stated reason for this abrupt u-turn is that AI agents changed the math around cross-platform development. Shopify’s argument is that AI has reduced the cost of building and maintaining the same functionality twice so much, that the benefits of shared implementation no longer outweigh React Native’s intrinsic drawbacks.
In this article I want to understand what AI actually changes in the native vs cross-platform equation, how much Shopify’s experience proves that maintaining two standalone apps became much more attractive, and whether their conclusion can be generalized to an universal rule of thumb.
AI benefits the native apps, but also mitigates many of cross-platform limitations
Now let’s take a deeper look at the main Shopify’s argument, that AI reduces the relative advantage of cross-platform mobile frameworks compared to the native development. On the first sight, this idea seems reasonable because we all feel that AI makes building native mobile apps simpler. What this argument misses, though, is that AI also eliminates many of the historical challenges of cross-platform development.
Take, for example, the case when a new native capability appears and you want to use it from React Native from day one. In the past, you’d have to learn how to write a native module or component that exposes this functionality to React Native, or you’d be stuck until someone publishes a library that does it. That’s not a trivial task, and it breaks the React Native abstraction by requiring a decent level of understanding of the underlying native platform.
Since this scenario can happen on both iOS and Android, in the past, you’d have to be versed in three tech stacks to have immediate access to all the new native APIs. Today, AI can generate the necessary wrappers from the platform API documentation. I know this firsthand because my own cross-platform app uses native maps UI and native maps libraries on both platforms, yet I have almost zero knowledge of how the iOS code works. I just declared the abstractions, and AI took care of the iOS implementation.
The above is just one example out of many: platform APIs, build systems, tricky interoperability issues and many more aspects, which were the weak spots of cross-platform development for years, became dramatically simpler with AI.
Therefore, while Shopify’s article is right that AI mitigates one of native development’s biggest disadvantages, their argument ignores the fact that AI also mitigates many of the disadvantages associated with cross-platform frameworks. It’s far from obvious which side benefits more.
Maintaining parity between two native applications is still expensive
Shopify’s article argues that:
[AI] Dramatically reduced the cost of maintaining parity between platforms through shared specifications, tests, and review checkpoints
That’s a very strong statement from a company that started maintaining fully native mobile apps under the new AI regime just several months ago.
From my experience, maintaining parity between complex native apps in the longer run is still a big challenge. I guess it can help if you rebuild the apps from scratch with the goal of leveraging AI in mind, but this adds costs by itself. Shopify is replacing shared implementation with specifications, tests, specialized tooling, review checkpoints and release processes. Building and maintaining all of that takes much effort too. There will also be recurring AI costs on top of that.
What Shopify did demonstrate is that AI can accelerate a migration when agents have an existing application as a reference implementation. They haven’t demonstrated yet that the same level of benefits remain when maintaining two independent implementations over many years.
Therefore, the most generous interpretation of this statement I can muster is that Shopify hopes that AI will greatly reduce the tax of maintaining parity between two standalone native applications, and that the initial results seem promising. That’s something, but the jury is still out whether the long-term absolute and relative reduction in effort (whether human or AI) changes the cross-platform trade-off in any way.
The AI angle of Shopify migration back to native mobile apps
Said all the above, I do think that AI contributed to the Shopify’s decision to sunset React Native.
See, in the past, when companies faced the consequences of adventurous tech bets, their hands were often tied. Refactoring entire applications from one technology to another is a very expensive and risky endeavor, so it wasn’t a viable path in many cases.
Nowadays, AI is indeed very good at re-implementing existing products in different tech stacks, so the cost of course-correction became considerably lower. In this regard, AI might be a true game-changer by helping many organizations get free from their unfortunate or no-longer-beneficial past tech decisions.
There were hints that React Native isn’t good fit for Shopify from the start
When Shopify announced that they’re moving to React Native back in 2020, my first reaction was to wonder what folks at Shopify believe they know that Airbnb didn’t.
See, AirBnb had decided to adopt React Native way back in 2016, and then went back to fully native apps in 2018. To explain their decision, Airbnb published very insightful five-part series of articles explaining why React Native didn’t work for them. So, when Shopify announced their desire to go all-in on React Native in 2020, in a post that so closely echoed Airbnb’s articles from 2016, it raised my eyebrows.
There are many caveats here: several years apart, different companies, etc.. However, if you read the Airbnb’s posts, you realize that they adopted React Native successfully, and developers were largely happy about it, but this experiment still didn’t work out in the grand scheme of things. One could very reasonably predict that Shopify might run into similar problems down the road.
Therefore, I can’t rule out the possibility that “we’re migrating back to native due to AI” is somewhat similar to an overstaffed company announcing “we’re laying off employees due to AI” – AI can serve as a catalyst and a convenient narrative rather than the actual reason for the change.
Cross-platform mobile development is the future
Taking a step back from Shopify’s specific situation, let’s discuss the cross-platform mobile development in general.
As I wrote in January 2025, I believe that, if you need to target both Android and iOS, it doesn’t make sense to write two completely separate applications anymore. React Native, Flutter and Kotlin Multiplatform are great frameworks, so it’s mostly just the question of what framework to choose.
I’m a huge proponent of Kotlin Multiplatform, but I think that React Native has its place as well. Many types of businesses can leverage this technology to achieve amazing results, in less demanding domains or supportive functions. Just don’t choose React Native when you build mission-critical e-commerce apps at an extremely successful and quickly growing multi-billion corporation. In my opinion, it simply doesn’t fit this use case well, and never did.
Speaking of Kotlin Mutliplatform, I think it could be an ideal choice for Shopify at this moment: keep the UI and platform-specific functionality fully native, while sharing networking, data access and other business logic between Android and iOS. Shopify experimented with KMP years ago and reported positive results, so I’d be very interested to understand why it isn’t part of their new strategy.
As usual, thank you for reading and feel free to leave comments below.
