Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Report: Apple's AI Strategy Could Finally Pay Off in 2026
- Thread starter MacRumors
- Start date
- Sort by reaction score
Fantastic? Pardon?
I have it enabled for CarPlay and it's mediocre, at best.
100%. They (Nvidia) was defending themselves the other day saying they "aren't Enron". Investors see what is going on and are asking a lot of questions.
NVidia and OpenAI, at the moment:
The maze on the milk package is unsolvable.
Oh, and the milk is both “whole” and “lowfat”.
Last edited:
I've been developing apps on Apple platforms since 2013, so yeah I totally noticed. They went from Vision Pro to AI at break neck speed. Like I said in my post, I believe this was the mistake. They got hyped, just like everyone else, and when push came to shove saw they couldn't do it in an 'Apple way'.
No one is profitable at AI, AI is generating slop, the proliferation of 'agents' shows that you can't just train larger and larger models. Instead of Apple doubling down on a tech they knew wouldn't be either economically effective or deliver after the hype, they pulled back. They're smart to pay for an API as opposed to investing billions into their own stack at this point.
I agree but the ecosystem does have a polish to it you can't find anywhere else. Like I can't see how LLMs use your person data now (varied levels of access, sometimes mistakes writing, not keeping context across things, like even Claude projects, etc.) I think they'll want like an AirPlay level of integration - Siri will know what you've worked on in Pages, how you talk about it in Messages, etc.
I know how LLMs work. I used Siri for a long time before turning it off. I am not interested in either. I don't need a 70% correct half wit to argue with. I need determinism. That's what I am using a damn machine for - to do the things humans aren't good at: repeatable, controlled determinism.
I think this is the correct way of looking at it.
If it reminds me of anything, it’s how Apple, in 2010, tried to respond to the social media boom of the 2000s by releasing Game Center and Ping in short succession… neither which really went anywhere.
Apple then realized that maybe social media platforms aren’t their strong suit.
The same thing as happening here. They tried to respond to a boom of a new type of Internet service, they didn’t do a particularly good job, but it ended up not mattering in the long run because (outside of Siri still desperately needing dramatic improvement) no one really wants to mess around with AI services, especially from Apple.
Apple is a hardware company that happens to make their own software. But they don’t do particularly well is Internet services. They have never replaced Google or any search engine for that matter, they don’t have a YouTube equivalent, they’ve given up on social media, AI is just the next thing.
I think another thing people are missing the point on is the M5 chips. A Mac Studio with an M5 Ultra chip will be anywhere between 20 - 35% of an Nvidia Blackwell based chipset, with two major things to realize:
- Nvidia GPUs don't run with unified memory the way Apple M chips do
- Mac Studio Ultras are a fraction of the price of enterprise grade Blackwell cards
- For the same price you can stack Mac Studio M5 ultras and start equaling performance using a lot less energy
When Apple does this their way, they'll own the complete stack, not be subjected to downstream market effects and offer a more integrated solution for users on Apple platforms.
- Nvidia GPUs don't run with unified memory the way Apple M chips do
- Mac Studio Ultras are a fraction of the price of enterprise grade Blackwell cards
- For the same price you can stack Mac Studio M5 ultras and start equaling performance using a lot less energy
When Apple does this their way, they'll own the complete stack, not be subjected to downstream market effects and offer a more integrated solution for users on Apple platforms.
Feels like we've been talking about Apple AI for three years, and Siri improvements for a decade. But nothing has happened. Let's just say I don't have faith that anything will change.
AI in 2031. Maybe.
AI in 2031. Maybe.
This metaphor only makes sense if Apple is trying to outrun a bear (other AI companies) which they absolutely were not, even from the start.
Even as far back as June 2024, Craig Federighi was openly saying to anyone who would listen that “Apple is *not* making a chat bot. They are not developing SiriGPT”.
He literally said that, before any of the delays, before any of the problems were publicly known, before anything.
Obviously, they were attempting to use large language models to complete some tasks within their software that they were not able to, but they were not competing with Gemini. They were not competing with OpenAI.
Meanwhile, you know where Apple is focused? Do you know where they are dramatically ahead of OpenAI by leaps and bounds?
The one thing they have been good at since 1976, hardware with in-house developed software.
Meanwhile, OpenAI is promising that they will have some type of hardware product…uh, um… some day? I think the latest reports are they hope to have their product out by late 2027? Or maybe 2028? So, you know, when we are on the iPhone 21? That’s a pretty long wait for something that is not guaranteed to be a success.
Yes old Siri used phrase pattern matching. It's basically voice controlled Zork on steroids.
But let's be careful when we say people want an LLM powered Siri. More carefully described, they want the LLM-powered Siri that was promised. Constraints on the technology, both technical and financial, will never deliver that Siri. Every single day that becomes more evident.
If we strip this back from the "hit everything with the LLM hammer" technology fad, what people really want is a better interface for their technology. They screwed up the already inadequate one quite seriously (liquid glass).
Nothing has improved anywhere in the last 3 years.
And with the M5 Macs, OLED iPads and the record setting iPhone 17 series, it’s hard to argue against the fact that their actual products have never been better.
Their software has been better… but it is also been so, so much worse. Looking at you iPhoneOS 2.0, iOS 4, iOS 7, Mac OS X Chita, Puma, Leopard, Yosemite.
I think the article is giving Apple too much credit. I just think they haven’t been innovative enough in AI. We have 14 years of Siri to prove it.
But I guess not being aggressive in this space will benefit them in the end when the AI bubble bursts.
But I guess not being aggressive in this space will benefit them in the end when the AI bubble bursts.
Then all of those Anthropic lawsuits become apples lawsuits, I wouldn’t take that deal.
Likely the same reason they didn’t buy perplexity, even though they were looking into it.
Flailing? Hardly.
Apple has been leading in privacy preserving AI, and developers have been adopting the Apple Foundation Models and using them to provide useful and practical features in their applications. Not just chatbots, but actual useful features, like when TripIt added a feature that can reprocess unfiled emails that their web services weren’t pre-configured to handle. A bug pain point for business travelers solved with the ondevice AI without the developer having to pay anything, and without them having to supply and download a model either.
Apple’s AI researchers have also been doing some very interesting work around the use of models specifically trained for use in wearable devices that can identify medical problems just from the sensors already on a watch.
It’s foolish to turn off a significant feature in your device merely because you misunderstand what it does or what it’s for.
Anthropic has nothing useful, and is antagonistic to Apple’s user-centric, privacy protecting strategy.
I used to be Team Android. I loved that Android was always first to market with all the cool new toys, and laughed at my Apple friends who couldn't play with them for another couple of years until Apple finally got around to matching them.
Then I realized that when Apple DID finally release a similar feature, it was well thought out, fully developed, and ready to go. Unlike Android's implementations that were often rushed and only half-implemented.
I've been team Apple for about 5 years now. I have every faith that their implementation of AI features will be much more useful than most of the "toys" added to Android at this point.
Then I realized that when Apple DID finally release a similar feature, it was well thought out, fully developed, and ready to go. Unlike Android's implementations that were often rushed and only half-implemented.
I've been team Apple for about 5 years now. I have every faith that their implementation of AI features will be much more useful than most of the "toys" added to Android at this point.
Apple is focused on AI features that deliver useful capabilities to users. Not gimmicky chatbots that nobody uses correctly anyway.
I don't buy it. He says it was painted by Keith Thompson. They asked Keith Thompson and he said "I don't comment on works that I was commissioned to do." If it was AI that someone said was him, surely he would have said "uh, no I didn't paint that."Tim Cook posting AI slop to promote Apple TV is pretty embarrassing.
Tim Cook Posts AI Slop in Christmas Message on Twitter/X, Ostensibly to Promote ‘Pluribus’
Link to: https://x.com/tim_cook/status/2003873311504130173daringfireball.net
I am sick and tired of asking my HomePod about something and that thing just can’t answer. Imagine when they update HomePods: current generation will become obsolete instantly. So I hate Apple for that. I need Siri that understands what I say and push informations I want
