It’s far too early to say anyone has fallen to far behind. Nothing about LLM based technology is a good business moat.
FWIW the current crop of AI enabled browsers are walking security problems. Think prompt injection attacks. Doing this well will take a long time.
I agree, and especially when it comes to Apple. They are not in the same business as OpenAI, Microsoft, and Google, etc. And, AFAIK, they have no plans to develop any major business systems or data centers large enough to rival MS and Google.
IMO, Apple just needs to develop a new digital assistant that’s good enough to get people to upgrade their iPhones more often (cause “significant iPhone sales growth”). And they probably have another year before the market starts having doubts about their ability to do that.
In the meantime, I’d guess most businesses will continue to choose the hardware, systems, and software their employees are allowed to use.
Blackberry also didn’t have the massive ecosystem and other lock-in effects that Apple did. They had some, but not nearly as much.
This is at least partially true, but at this point of the game AI is more platform-agnostic. I think the runway between “where we’re at now” and “AI operating system that makes Microsoft Word obsolete” is pretty long, and gives Apple much more time to catch up.
Remember, developing their own AI is only one way for them to get where they’re going. They could buy an AI company. They could license third-party AIs that get better. They could offer operating system hooks that allow a user to integrate whatever AI they want. Some of those options wouldn’t be available to a smaller company, but Apple has some pretty serious resources to throw behind their AI efforts.
What does it mean to be in Apple’s ecosystem? Is it using exclusively Apple devices, apps, and services? If so, I’ve never been in Apple’s ecosystem, and I don’t know anyone—young or old—who is.
My desktop computer, my phone, and my tablet are all Apple devices, but I’ve never used anything but a hybrid app stack, and I can’t imagine relying exclusively on Apple’s apps and services—most of them simply aren’t suited to my needs or my tastes. Nothing Apple does or doesn’t do with AI is going to lure me away from the third-party apps and services I use.
A lot of people end up in whatever ecosystem their employers use. Between them, Google and Microsoft have a lock on the enterprise office stack; I suspect that nothing Apple does or doesn’t do with AI will change that, either.
My definition is using primarily Apple devices and many of Apple’s services, but not exclusively. For example, I have every Apple device (I think), and more than one for some, such as HomePods, iPads, and Apple TVs. I also have an Apple One subscription, which means I get a considerable amount of news from Apple News+, use Apple Music exclusively, and so forth.
However, I also use Google products and occasionally Microsoft Office because the school uses both of those platforms in addition to being a one-to-one iPad school. I also use Ulysses (book project) and other apps and services not provided by Apple.
My larger point is that I am not confident that Apple’s hardware alone will maintain a low churn rate if AI becomes a ubiquitous computing platform and service that goes beyond the typical consumer chatbots and Apple fails to deliver a comparable quality of service for the types of customers they primarily serve. I’m suggesting that we may be in the early stages of a computing paradigm shift with AI.
But, as I said above, “I readily admit no expertise in this area … This is the larger threat I see as a complete amateur in such matters.” 
This is what I’m trying to figure out. What do you mean by the above?
In the mobile arena, I see two ways this could fly.
The first is that there would be an AI company that actually started producing an operating system and hardware to run it. In that case, Apple and Android would both have some challenges.
The second is that there would be some sort of app/website that would be ubiquitous, that would facilitate access to the AI services. That would almost certainly run on Android/iOS devices.
Are you thinking one of those two scenarios, or some third option?
I may not be thinking well at all! 
What I have in mind is—for lack of a better way to describe it—a hardware-agnostic operating system that combines AI, AI agents, and related technologies into a unified computing platform. This system would integrate file management (including conversions, compression, and sharing), an AI-infused browser, and agents capable of writing text, converting it to compatible legacy formats, creating spreadsheets, transcribing meeting notes, managing projects and tasks, scheduling appointments, drafting slides, and more. In other words, a computing platform that merges operating system and application functions to accomplish most of what we need.
This may be far off, but OpenAI,if one can believe them, a significant qualification, is predicting “a legitimate AI researcher” by 2028, and I just had Claude create reminders for me. AI will also convert my files to different formats and, as we know, much more. Currently, those are discrete features or actions. What if all of it could be melded together into one computing platform?
I have no idea if such a thing is possible. I know there are many complications and unintended consequences, but in theory, I do not see why a sophisticated AI platform could not be developed that could do much of what I describe.
What the heck are you talking about? All we have now are large language models and machine learning. What makes you think those two things can lead to something that is actually intelligent?
I think you, and many others, are a victim of marketing hype and the pervasiveness of pie in the sky science fiction movies and books.
I probably am an expert in this area, mostly by accident. Out of frustration, with the hype in the past year, I’ve learned a lot about LLMs. The current generation of technology can be replicated by people willing to through enough GPU processing power at the problem. Witness DeepSeek early in the year releasing an open-source model that was competitive with the closed-source frontier models.
Given the core approach to building LLMs is now well understood, we can even play games like distillation, I can’t see how OpenAI or Anthropic et al have business moat.
The best either of them have is mindshare, that is a start, but not enough to takeout Apple, Google, Microsoft, on it’s own.
In fact, I think if Apple makes the right play, access allows Gemini or Sonnet to access our on device data with the right privacy protections and Apple is ahead of the others without having built a frontier model of their own.
I think this is the web, no? I could probably replace 95% of the apps on my Mac with either a web app or a web interface without materially hampering my ability to get stuff done if I were comfortable working online and in the cloud. It’s not that complicated to build a web-only workflow that incorporates AI and I’m not convinced that putting all in one interface would make it any easier.
We’re already there.
The web itself is no longer a hardware-agnostic operating system, since Google keeps trying to make it their own platform in lieu of not having a hardware platform of their own like Apple and Microsoft do.
True, but it’s still agnostic enough to work with. Are you thinking of some Alphabet-driven protocol that hobbles our ability to use the web as we might wish?
Google leads in setting so-called web standards. Most corporate software that I am familiar with “runs best” on the Google Chrome browser.
With respect, (and it is genuine) if you’ve read my prior posts, I’m a critic of the AI hype. 
I’m not a believer in AGI or ASI. That said, I can foresee a new computing platform based on AI capabilities, but as I caveated several times, I claim no expertise in these matters, I’m just considering possibilities.
For example, Google AMP: how Google tried to fix the web by taking it over | The Verge
EDIT TO ADD: Since the Verge article is now behind a paywall, here’s another link.
Google AMP is Dead: How to Optimize Your Site for Speed and Google Rankings
Is Google Being Evil Again With the Launch of AMP for Email?
I wish Apple would allow Chromium browsers on the iPad and iPhone in the US, like they do in the EU. I still find sites that don’t work correctly with webkit.
That’s because Google keeps promulgating new “web standards” that benefit their web-oriented business model.
And Apple can’t be bothered to adhere to those standards?
Google released Chrome in 2008 and made Chromium open source at the same time. Chrome became very popular very quickly and eventually so did Chromium. Today, as far as I know, Safari and Firefox are the only major browsers not based on Chromium.
I am not sure Safari needs to become an AI browser. If the promised context aware Siri with a bit more smarts arrives it could go a long way for privacy conscious agentic AI.
@Bmosbacker your described ubiquitous computing platform sounds a lot like Google Workspace or MS 365.
Read this very interesting thread which is an example of what I’m talking about.
For those who are “all in” with Google, whatever Google pushes is a standard. Not everyone agrees.
Jen Simmons: “Imagine one group advocates …” - Front-End Social
Imagine one group advocates for a new web technology, and write down their ideas. And then other people have concerns about those ideas (maybe thinking it’s not well designed, not easy for developers to use, invades privacy, isn’t secure, etc)… so they are totally not on board. If those ideas end up in a web standards group anyway, and get pushed through while the original group still objects — is that a web standard or not? If a browser formally objects & refuses to ship, is it a web standard?