Kuo: Apple's AI Deal With Google Is Temporary and Buys It Time

[bluecoast]

Google managed to deploy both self-driving cars and an LLM in the same timeframe where Apple spent billions of dollars on its own initiatives in those categories with nothing to show for all its outlays.

You've hit the bulls eye.

Like you, I find it barely believable that Apple solved the same problems as Waymo did and continues to do, but then shelved it without having anything to show for it.

And again, I find it barely believable that Apple will be able to meaningfully compete with Google with frontier AI models.

Apple's strength is not at all in AI, but in ML as it applies to their hardware (cameras essentially).

Apple may just have quite limited goals i.e. make a great personal assistant for the iPhone, Mac etc.

But given that frontier AI models will unlock things like robotics, understanding the human and natural environment etc. just making a great personal assistant is going to limit them in the future. And any glasses product will need the latter.

It's not even got a strength in cloud computing, which for google is a complex compute platform.

Apple uses it as a CDN, storage and sync platform i.e. about the most basic things that that cloud does nowadays.

And by all accounts, Apple has barely any owned cloud capacity, renting most of it. Which is sensible, given its scale, but they don't just seem to have the same expertise that Google has.

I think that Apple has had an amazing run with the iPhone and now the Mac is great again. I love both of these products.

But I think that they're not at all set up to compete well with the next era of computing. They're struggling and will continue to struggle.

If LLMs had not come along, Apple would be in an incredibly strong position. But Cook has mostly optimised them to keep pumping out iPhones and the Mac etc. and is has not at all well equipped Apple for the upset that chatGPT was.

At least they are making huge amounts of money for acquisitions which they will have to do.

Or I guess they could merge with another very complementary company that is set up for the next paradigm of computing i.e. Google.

 

[Andres Cantu]

They need to be careful though because they will be setting expectations once the Google one is used by users. Apple’s won’t be able to be of lesser quality once it launches.

 

[bluecoast]

Buy time until the AI fad passes and the bubble pops. Save some money and use someone else’s AI.

Agreed, but the ownership of the current frontier models is quite complex. ChatGPT/OpenAI is very arcane and Microsoft own a large proportion of the commercial part of OpenAI. Hard to say how they will let apple buy this off them.

Anthropic I believe are in a similar position to ChatGPT, but with Amazon being their primary benefactor. Again, hard to see that Amazon would let Apple have them.

Mistral and Le Chat - not sure that the EU would let any US company buy them.

Facebook are hardly likely to sell off anything that they are doing to Apple.

Then who else? Lots of other companies who are riding off the back of these frontier models.

The only thing that I can think that will save Apple is it can go on a huge talent aquisition hiring spree when the bubble does burst. But when will it?

In all that time waiting for the pop, Gemini is just going to get better and better. Ditto ChatGPT. Hard to see how Apple can catch up here with anything that they are making internally.

 

[WarmWinterHat]

...

Or I guess they could merge with another very complementary company that is set up for the next paradigm of computing i.e. Google.

I don't think there is a country on the planet, including the US, that would approve that merger.

 

[Yr Blues]

The frontier models don't hallucinate 40% of the time and if you think that it's very obvious your "AI sucks and will get worse" statement is completely unsupported and makes zero sense in the context of how quickly AI is improving.


Also, temporary? Until Apple can catch up to Google's AI capabilities? So...permanent.

How many shares did you buy. 🤣

 

[bluecoast]

I don't think there is a country on the planet, including the US, that would approve that merger.

Never say never.

10-15 years ago, no one would've ever thought that the US government would have to bail Intel out.

Tech is ruthless isn't it? Today's winners can rapidly become tomorrow's has beens.

Just ask RIM (Blackberry) and Nokia (handsets).

 

[WarmWinterHat]

Never say never.

10-15 years ago, no one would've ever thought that the US government would have to bail Intel out.

Tech is ruthless isn't it? Today's winners can rapidly become tomorrow's has beens.

Just ask RIM (Blackberry) and Nokia (handsets).

They didn't have to. They chose to.

 

[AppleFan91]

This isn’t surprising. Apple was in a tough spot because they missed with the first Apple Intelligence rollout and people are expecting “Siri 2.0” this spring. This allows them to roll out “Siri 2.0” while also working on their in-house solution without a timeline and pressure.

They’ve done this throughout their history. Partner with an industry leader while working behind the scenes on your own solution to eventually bring it in-house. (Google Maps, Weather Channel, TomTom, Intel, Qualcomm, etc). This is temporary, though may last half a decade or something, but they will eventually bring it all in house like they usually do.

 

[Yr Blues]

Lemme guess. People just want Siri to sound realistic and attractive. 😆

 

[arthurgd3]

This isn’t surprising. Apple was in a tough spot because they missed with the first Apple Intelligence rollout and people are expecting “Siri 2.0” this spring. This allows them to roll out “Siri 2.0” while also working on their in-house solution without a timeline and pressure.

They’ve done this throughout their history. Partner with an industry leader while working behind the scenes on your own solution to eventually bring it in-house. (Google Maps, Weather Channel, TomTom, Intel, Qualcomm, etc). This is temporary, though may last half a decade or something, but they will eventually bring it all in house like they usually do.

You're previous examples of Apple doing this before are valid but absolutely not comparable to the AI conversation in scope and scale. I'm not doubting Apple can do it's own thing behind the scenes for the next half decade but no one should be underestimating how much further along Google's AI will be in this same timeframe.

 

[dba415]

Google managed to deploy both self-driving cars and an LLM in the same timeframe where Apple spent billions of dollars on its own initiatives in those categories with nothing to show for all its outlays.

While also releasing a cell phone, improving Android, and innovating quantum research.

Google has about the same number of employees at Apple too. What the heck is Apple doing?

 

[mdriftmeyer]

It's rather obvious this is temporary as their server AI cabinets with M Server series are fully tested in-house, then built out around the globe.

 

[MacMorrison]

That's a critical mass of potential privacy violations - a deal with the devil.

 

[mdriftmeyer]

Buy time until the AI fad passes and the bubble pops. Save some money and use someone else’s AI.

AI will evolve into backend services and heavy HPC usage across all frontiers of applied sciences.

 

[bluecoast]

They didn't have to. They chose to.

OK. But I contend that my overall point still stands.

Make a major wrong strategic call in tech and then keep on digging and obscurity beckons.

And/or suddenly getting exposed because someone else changes the world - instead of you.

I would say that Apple is in a similar situation now with LLMs/AI as its competitors were when it launched the iPhone.

It wasn't just the amazing hardware and design that destroyed RIM and Nokia - it was coming up with an incredible platform phoneOS / iOS from the start that could then run apps.

RIM and Nokia had nothing like this, with barebones operating systems that did phone things with email tacked on. They just couldn't compete.

I would contend that the tech around creating LLMs is so alien to what Apple does that it's going to find it extremely hard to create its own frontier platform that's competitive. It would take a massive change at Apple. And meanwhile, Google Gemini and ChatGPT will just keep on getting better each month.

Let's hope that Cook stands down soon as I don't think he's the person to try and fix the situation, given he's lead Apple to fall flat on its face with AI so far.

 

[Hogswarts]

As long as Apple adheres to its privacy principles and practices, I'll remain a customer. Even the best AI available still sucks. I appreciate their effort to bring the best to their customers.

 

[apple2022]

Apple is starting to sound like those legacy car manufacturers that are planning on getting into autonomy “in a few years”
lol

 

[TechRemarker]

gemini is easily the best ai model out there. gonna be hard for apple to create something that tops the best.

As someone who actively uses Gemini and ChatGPT, Gemini is a far cry behind ChatGPT is real world use. Have done extensive testing of complex questions between them both and Gemini has consistently fallen short for involved requests. But I'm sure Apple doesn't want to choose ChatGPT any more than they have to, since as ChatGPT gets into hardware, they have the potential of being a major competitor to Apple. Vs Google who has of course much better AI than Apple currently but has been terrible at hardware compared to Apple, so I assume Apple is much less worried about Google being a competitor. It's unfortunate Apple has had so much time to build the best Siri and every year modest to non existent improvements. And if not for all this LLM progress, Siri I'm sure would be the same today as it was mostly 5 or more years ago.

 

[jz0309]

I don’t see any contradiction here.

well, TSMC's capacity at high-end nodes is limited.
Apple's chip volume will be increasing
Adding "server" chips requires capacity

maybe Apple is looking closer at Intel than we think

 

[switz]

Ford says "we almost have the Edsel ready for mass production...." but it was a dud. Apple says "we almost have AI ready..." wonder if it too will be a dud? Both were leaders in their respective industries.

 

[jz0309]

TSMC is probably looking to diversify and profit from the AI hype, can you fault them. Contracts with Apple still exist it may mean adding extra production or a negotiable tactic for Apple to pay more.

I think it is more likely that Apple is the one trying to diversify ... TSMC has more to loose as Apple is a steady ever increasing volume, AI is a bubble, profitable for a foundry though

 

[arthurgd3]

As someone who actively uses Gemini and ChatGPT, Gemini is a far cry behind ChatGPT is real world use. Have done extensive testing of complex questions between them both and Gemini has consistently fallen short for involved requests. But I'm sure Apple doesn't want to choose ChatGPT any more than they have to, since as ChatGPT gets into hardware, they have the potential of being a major competitor to Apple. Vs Google who has of course much better AI than Apple currently but has been terrible at hardware compared to Apple, so I assume Apple is much less worried about Google being a competitor. It's unfortunate Apple has had so much time to build the best Siri and every year modest to non existent improvements. And if not for all this LLM progress, Siri I'm sure would be the same today as it was mostly 5 or more years ago.

Just out of curiousity what kind of complex questions are you asking Gemini in real world use where it's falling short and I'm not sure what you mean by "involved requests".

 

[ethanwa79]

gemini is easily the best ai model out there.

HARD disagree. Anthropic Opus 4.5 kills Gemini when it comes to coding, for example, and I find ChatGPT 5.2 to be better than Gemini in a lot of ways. Google does have the lead with image generation though with Nano Banana Pro.

 

[zarmanto]

Nothing is more permanent than a temporary solution.

(In case it isn't obvious... yes, this road is paved and appears well maintained.)

 

[FyrStrike]

I personally don't want anything to do with Google on my iPhone. I hope we can turn it off. And if this is short term I hope either Apple makes their own LLM model or at least gives us the choice to choose which AI platform runs AI within our devices. Im a ChatGPT dude. Gemini gives me the "heebeegeebees" thanks to Google's spy practices which I prefer to minimize.