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.
Apple's Q3 2024 Earnings Call Takeaways
- Thread starter MacRumors
- Start date
- Sort by reaction score
lol wait couple days for this to turn out be a dud. Let’s pull up the hype of Gemini Pro in last iteration. Google had to admit the demo was faked/fudged, some of the benchmarks didn’t add up. I had 2 month free trail, which gave me 30 TB free data along with Google business goodies. Gemini pro was the worst, despite the fudged demos and hype. Some of the actual inventors of transformer models, LLM neural nets have a very different take if you get past the fake demos and gamed synthetic benchmarks. The game hasn’t begun, it’s way too early.
And on device models have beat even GPt4 models for contextual AI. A 3-4 Bit quantized model with right access to on device data can be much better than a generic large model. I would be surprised if Apple wants to play the whole game. Provide on device models, hybrid in Apple data center and leverage Open AI for any generic tasks.
Last edited:
Of course they wait. Except when they don't. Like GUI PCs (Lisa, Mac). Or smartphones (iPhone bore no resemblance to anything that came before it). Of course, we know those product categories are exceptions that had no impact on the market. Oh, umm, wait...
Amazingly terrible take. AI has barely begun, and Apple has had their own models for years - a number of which are published and open source. It is also far from clear how much of a lead NV has on Apple's chips, if you're Apple and understand fully how to use them and can build servers to optimize whatever you want (RAM size and bandwidth, perhaps). We don't know what's going into Apple's DC to support PCC, but I'll bet it's not just stock M2 Ultras.
NV has gotten ridiculously rich off an utterly unsustainable bubble. They are this century's winner of the "greater fool" game, so far. This is not an indictment of their product! It's an indictment of all the idiots spending billions to use them for services that can never be profitable. And that doesn't mean that *nobody* will make money of AI. But most people won't. (My money is on Apple being one of the companies that will make money, BTW.)
Many false assumptions and premises here. The biggest one is that the course of AI development is set. You are assuming you know how AI will improve over the next few years. You don't, and likely nobody does.
Don’t believe all you read, MS is just putting that in the documents to appease future fed regulators and such.
And I agree with you that the stuff Apple is doing with AI at the OS level is interesting. But honestly they should have seen this coming and have a new 100% AI Based OS already built or at least in testing. It’s going to be a complete game changer more than the last OS rewrite.
You don’t have to believe in anything. Just look at Microsoft’s own models, their actions and their dependence on Open AI. Microsoft pushed for partnership with Mistral, earlier this year because they knew Open AI could be potential competitor. Like Steve Jobs once said, Microsoft is opportunistic. If it was for appeasing regulators, they wouldn’t have gone out looking for partnership with mistral. It’s too early in the AI game to announce winners or losers.
I have to stop you when you argue that AI is an unsustainable bubble. It is the future full stop. You sound like Balmer when he laughed sitting at his desk with the first brand new iPhone in his hand.
Apple is being left in the dust. If they had excellent AI chips and servers designs they would be building them and selling them because they would be worth more than SUVs each not hiding them in files on their computers.
You are right with that. But Apple should be much further ahead than they are. That’s all I’m really saying. They can’t even get out a decent release of Apple Intelligence Version One without delays.
its like declaring marathon winners and losers after 500 yards. Microsoft knows they are not ahead, heck if you look at Open AI’s own scientists, they are very cautious. It also depends on the strategy of Apple, all I know at this point is Apple has a very different approach with on device inference. Apple has released lot of open source models and frameworks in past couple of years. Apple’s own Quantized 3.5 bit model beats Microsoft’s Phi mini and llama 3 8B.
I have been trying writing tools /apple intelligence on my iPad Pro, it does a great job of summarizing, key points. I was using llama 3 8B on MBP for similar tasks. Apple Intelligence so far does a better job, that matches with what Apple published. Creating a table from text is broken in the Beta 1, It’s an early beta version, so it has bugs and hopefully Apple can fix it soon. Math notes in calculator App is another favorite of mine, just draw sketches and free hand math, it does a good job of analyzing and providing solutions. It doesn’t have to be ChatGPT models running on cluster of servers all the time. Summerize my emails, tasks, reminders and alert when something needs attention. It’s just beginning.
They need to have a look at their pricing on the Mac, nearly all of the models are great machines until you see the price.
What I don't understand is why so many people are focused on what are just glorified chat bots. Apple's implementation of AI is much more likely to be the way of the future. Once the usual IT Hype Bubble settles down, and AI has evolved, it will mostly be a bunch of tool and utilities that use LLMs as part of how they function. Chat bots are largely a gimmick.
As far as I understand it, Apple has made conditions for third party app store so onerous that it is worth it for almost no devs. Which is why they're back in court
I love this quote, and I think it speaks volumes about the overall shortsightedness of Macrumours members. They are overexaggerating the lead the competition has, while totally underestimating the fact that Apple has very deep pockets, looks committed to this for the long haul, and retains control over their ecosystem.
It doesn’t have to be mutually exclusive. Like Steve Jobs once said, Microsoft doesn’t have to lose for Apple to win. Apple needs to keep up with device level stuff. There is a space for multiple winners at different levels.
"Even with Visoin Pro?" What exactly does it tell you?
Vision Pro, at this time and in the foreseeable future, is not expected to be any more than a little pebble in terms of revenue or volume for several years. There is zero expectation that it will. It is a product in development that can be purchased by the public. It is not even close to being a mass market product at this time. I highly doubt revenue from Vision Pro is a factor considered in analyzing the company's performance, especially since there is no way to know anything about Vision Pro sales/revenue.
Did you have an expectation that Vision Pro would be a positive or mitigating performance factor in the quarterly results?
You would see a decline across all their products. Its the ecosystem. Remove one product from the ecosystem and the whole thing becomes unstable.
Hey I hope Apple is able to catch up. I hope all their AI stuff is excellent. They have the customers locked in so that’s a plus. Hey I’m an investor and I am an investor in MS, Google, etc so I understand there can be more than one winner and it would be best if Apple probably stays safely in their lane.
But I do feel like Apple and Microsoft are missing the boat on these little AI features they are adding to the OS.
Are they thinking big enough?
How about an OS that is completely run by some sort of LLM? You want to run a x64 windows binary, sure it simply emulates everything you need for the processor, the Windows OS, the gpu library and runs the program, no CPU or windows OS needed. It literally recreates it all on the fly. Now you want to run a Commodore 64 game? Sure just run the binary and it will figure it out. It will even be able to decompile the code in real time and fix old bugs of software as it runs it.
There’s absolutely no limits what such an AI based OS could do. You could just describe an APP and it will build it and run it on the fly.
I know we aren’t there yet, but Apple is supposed to be 5+ years ahead in their labs, so let’s see a sneak peek of the true future of computing!
I didn't argue that. I argued that what's going on right now is a bubble. It's a ridiculous one and very soon - maybe in the next year, maybe in the next two, but soon - investors will lose tens of billions, as the last fool buys the last share at an inflated price and then looks around to find that nobody else is willing to pay for a fantasy with no path to profitability.
Great fortunes will be made on AI, but greater ones will be lost, at least in the short term. Over the longer term, I totally agree that it is the future. Or at least a significant part of it.
"Left in the dust" is, as I already said, a terrible take. And as someone else said, you're judging 500 yards into a marathon. It's just silly. Your argument shows that you don't understand the importance of ecosystems, nor of strategic moves. Apple can take or leave a few billion in profits from selling AI workstations (though they'll probably starting taking it, soon enough), but if they think they have a strategic imperative to build servers for their own data center, or allocate wafers to iPhone 16s instead of AI servers, that's what they're going to do.
The truth is, no victory in today's market is likely to matter much in five years, as everything will change. Probably a few times. As usual, they're taking the long view.
Says who? You have a published schedule that says there's a delay? At WWDC they said AI would be coming later in the year than the first betas.
It's done when it's done. Trying to push it out faster would be terminally stupid. I'm very glad they're not doing that.
Hahahahaha!!!
Are you serious?!? This is a ridiculous fantasy and if anyone tried to bring this up in an R&D meeting today they'd be laughed out of the room.
Someday, probably, there will be machines with enough power to do what you're describing. It won't be in the next ten years, though. Or, likely, the next twenty. And nobody will be stupid enough to do that even then - instead if necessary they'll have an AI design a purpose-built emulator, because it will run MANY orders of magnitude more efficiently.
There are interesting possibilities in AI making games or apps that... rhyme with older apps, much like history is said to rhyme but not repeat. But if you really want to run *actual* *specific* code, you're never going to tell your genius AI to do it for you directly.
I’m going to disagree on that, the Mac is a small part of the ecosystem. A lot of the younger generation don’t even realize that Apple makes computers and not just iPhones, iPads and watches. Might see a very small dip in services but that’s most likely it.
I think the truth is somewhere in between.
Mac is a significant part of Apple's revenues and earnings. "Significant" does not imply "majority", of course. But see what happens to your stock if you lose ~8% of your revenue!
Macs surely have less of an impact on Services revenue than iPhones do - the app store is a behemoth, and Mac is a fringe player there.
There are other ways Mac contributes, however. The dev environment for all other Apple products is on Mac, and while it *could* be ported to other environments, it has not been (and that seems unlikely to happen in the short and medium term). It also provides an internal customer for high-end silicon, which will trickle down eventually to other platforms (eventually, or perhaps right away, in the case of the iPad).
You don't drive a semi truck with 17 wheels. You don't want Apple to try to run without Macs either.
Outside Apple needs to catchup on AI bubble on MR. Many researchers who worked on original GPT 2 to GPT 4 are pretty execited about Apples different approach. Some scientists at Meta want Apple to incorporate their own models in to on device computing along with Apples own. Here is just one take by Andrej Karpathy, one of the early Open AI founding member till GPT4.