What exactly is 'holding out'? I'd be very surprised if any noticeable number of users switch away from iPhone to Google to use an app you can download on iPhone already. People switch to Android because it's cheaper, or because tech is a hobby for them and their phone doesn't have enough useless knobs to play with. If AI is going to be part of it, it's at minimum going to need to be exclusive to Android I would think.
I'm talking a bit longer arc than just an app or two (which, BTW, the Gemini app on iOS is nowhere near what Gemini is on an Android phone, not even close, it's as much that as a Pontiac Fiero with some bondo painted to look like a Ferrari is a Ferrari....but set that aside, it's not even about that).I'm fundamentally talking about two things:
1. How people interact with technology. The actual human-computer interface that to-date has been largely screen-based, direct interaction in one form or the other.
2. Openness of a platform _or_ the platform being _so damn good_ that openness doesn't matter.
Traditionally, and I think the point you're making, is that Apple does those two so well right now, it's not even close. I'd agree for the moment.
But, it remains to be seen if Siri's continued trainwreck is "Apple just being the best not the first" (which I think everyone outside of the PR dept at 1 Infinite Loop knows is a series of errors, personnel issues, mistakes, strategic errors, etc that have compounded that have not yet actually been solved entirely. They'll sort it out, no doubt, but how much of this lead do they continue to borrow from in the mean time is a question no one really has the answer to.
What makes that question more interesting (and the answer higher stakes) is how it relates with #2. Governments are putting more pressure on (and courts are upholding) the walls of the garden, so, that further weakens the position (and accelerates that timeline at some level), but that's a lot less important than, now, the stakes are *really* high for Apple to deliver in AI, wearables, developer platform (this one doesn't get the airtime it needs...its a big part of how Apple rose to dominance beyond the vacuum they created masterfully in the market, of course, but IDK if you've seen the app devex on iOS/macos right now, it's truly horrible...Swift is a hot mess and the tooling is terrible...it's bad when Python & JS ecosystems/toolchains are better by comparison, which used to be the opposite), vehicle integration, home assistants/devices, media consumption, etc.
The chinks in the armor will add up over time. They are adding up. Apple Intelligence not being great is a big problem. And the Gemini App on iOS isn't the full experience. Have you used a modern android flagship? If your assessment is "they're just cheaper with more useless knobs" I think that maybe sign you haven't touched one in a year or so at least. The Fold 7 sitting next to my iPhone 16 PM is a _far more_ seamless, low touch "just works" UX than iOS 18 or iOS 26 is (and I'm not talking reliability on iOS 26, it's a beta, I'm talking number-of-touches-to-do-something kinda "just works").
And that's the key thing: Apple's current product line is heavily dependent on your eyes on a screen, on a buzz in your pocket or on your wrist resulting in an interaction. It's also where they are best. The place they are weakest and have the least market presence, moat, differentiation, or leadership (they don't even play in some of these) are: home assistants/devices & appliances (Google's offerings are far and away more robust, useful, and better here, and a part of that is how terrible Siri is makes a HomePod an expensive and good sounding brick, but also, Google's is in far more form factors, from thermostats, to security cameras, refrigerators even!), cars (CarPlay is not good, Android Auto is passable, so this is anyone's game perhaps, but don't discount Google's investment in Waymo that is without a doubt the long arc where Apple's already admitted defeat: what happens when far fewer people are the actual driver in their car, but they're the Pax...have you been in a Waymo? You control the music and the ride itself from, amongst other places, Google Assistant and Gemini....you don't even need to touch a screen unless you're on iOS, b/c, well, Apple only wants you to talk to Siri), wearables (Apple does great here, the watch is truly great, and Android Wear was far behind for a long time, but again, Siri is the achilles' heel...Gemini on the new Wear kit is damn good), TVs (this one is scary for Apple, when you buy a TV that runs Android or the like, it becomes a lot harder to sell another $150 box to go with it...for now, Apple's winning on content, let's see how long they can keep that in their garden and if margins are going to be good enough here), MacOS/Computers (Windows is terrible, but heavily penetrated, good luck with using an iOS device to its fullest if you're on a Windows device, you won't, but you can with an Android device).
Also, Apple's most recent advancement they have wielded well as a weapon is Apple Silicon. That train is slowing down, and the Snapdragon and other advanced ARM designs are just as good in the mobile space now (not in tablet or laptop/desktop quite yet, but that's changing). Now, it's not about the silicon, _but what you do with it_, and there, man, Apple AI is *SO BAD*. The new Apple Foundation Models are *not good*. They are far worse than even the Chinese OSS SLMs of a generation or two ago. They are embarrassingly bad actually. Gemma 3n absolutely obliterates them. It's not even close. And that's Google's OSS offering *today* (not theoretical, not promised, shipping today and has been for some time). What good is industry leading silicon to run AI better than anyone else when your AI doesn't exist and app developers are bringing in Google's embedded model to run on your silicon? Arguably, the SoCs from Qualcomm will be better suited to run small LLMs and larger SLMs, 16GB on the newer ARM competitors will be far better to run larger models where on device compute is necessary (for Siri like interactions), Apple is only rumored to get up to 12GB with the next rev of the iPhone. That extra 4GB may not sound like much, but it's a LOT. It's the difference between a text only model and a multimodal one. It's quite possibly the difference between summarizing a news article correctly or in a way that suggests something totally false.
I do think visionOS has the most potential of any product in the lineup to keep Apple relevant and dominant. The Humane pin was nothing short of an enormous debacle/dumpster fire/failure, but the thesis is 100% spot on IMO. It'll be a few years before we see it "work", but Apple has to _really_ get its act together with AI to translate their current positioning & leadership into a space like that.
In a world where the screen isn't as important as it is today, Apple's advantage is seriously eroded and eroding faster every day they don't get Siri figured out, appliances/speakers that work well/a good portfolio, and make the native app great again (both in terms of needing one and writing/marketing one).
