Apple Caught Off Guard by AI Demand for Mac Mini and Mac Studio

[Wahlstrm]

0

That was posted before in this thread, also without giving any reasoning. There's nothing relevant the slots would add to a hypothetical new Mac Pro. Remember that the M2 did not allow for graphics cards in the slots and that memory could not be upgraded.

That kind of design wouldn't do anything for AI over the Mac Studio.

There used to be a rumour about an “M extreme” as the top tier above M Ultra.

Pre-AI that was probably not worth the effort for Apple, the Ultra was already overkill for 99%(?) of customers.

But now that Extreme would likely fly off the shelves, regardless of price.

Apple could also tier the RAM, like they do on most other devices.

MBA max out at 32, need more then buy a MBP. Mini max out at 64, need more then buy a studio etc. The Studio currently tops out at 256 with 512 being promised to come later.

Apple could keep 512 and perhaps 1024GB and 1536GB as an option for the Mac PRO.

For customers who’s workflow actually need these specs then the better cooling and perhaps the space to add a few dozen TB of storage for huge data sets or backups etc.

For some it would perhaps be nice to have the option of rack mount their “AI server”.

Redundancy in power supply, networking etc are also much appreciated amongst enterprise customers.

There’s plenty Apple could do with it.

 

[vantelimus]

I’m honestly not sure what point you’re trying to make here… Humans consume a lot of energy, living creatures in general consume a lot of energy. This isn’t exactly news. The humans NOT doing the job that the various LLMs can supposedly do more efficiently are still going to consume pretty much the exact same amount of energy. Ipso facto, my original point stands - Kill All Humans!

The point was simply that humans will consume 1000x more energy doing the work that can be done by an AI query.

What you say isn't true according to the data. A single desk-job worker consumes 133kWh/week according to the US Energy Information Administration. Average household electricity usage is 863 kWh/month according to the same agency. The American Community Survey of 2024 by the census bureau (census.gov), puts the average number of people per household at 2.5. There are 720 hours in 30 days. Doing the arithmetic yields 863/(720x2.5)=0.479 kWh per person per hour. Multiply that by 40 to get the equivalent of a work week and you have 0.479x40=19.16 kWh consumed at home vs. 133 kWh consumed by the same person at work.

 

[vantelimus]

So it is not the data centers per se taxing the grid, it is the agents of the slop creating and brain-work outsourcing humans! My bid to protect the grid: get rid of the agents, replace them with humans. 😊

The agent is several orders of magnitude faster and uses several orders of magnitude less energy from the grid.

 

[AppleTO]

There used to be a rumour about an “M extreme” as the top tier above M Ultra.

Pre-AI that was probably not worth the effort for Apple, the Ultra was already overkill for 99%(?) of customers.

But now that Extreme would likely fly off the shelves, regardless of price.

Apple could also tier the RAM, like they do on most other devices.

MBA max out at 32, need more then buy a MBP. Mini max out at 64, need more then buy a studio etc. The Studio currently tops out at 256 with 512 being promised to come later.

Apple could keep 512 and perhaps 1024GB and 1536GB as an option for the Mac PRO.

For customers who’s workflow actually need these specs then the better cooling and perhaps the space to add a few dozen TB of storage for huge data sets or backups etc.

For some it would perhaps be nice to have the option of rack mount their “AI server”.

Redundancy in power supply, networking etc are also much appreciated amongst enterprise customers.

There’s plenty Apple could do with it.

Something else they could’ve done is create some sort of Apple Silicon AI accelerator card for the slots. I’m sure those would have flown off the shelf as well.

 

[Jakol]

AI isn't going to go away, when it pops there will be just less money being tossed around.

Good, I'll take it as long as prices of RAM and SSD go down and nvidia suffers.

 

[Omega Mac]

It's interesting to see people still wanting the AI bubble to pop. Sure, it may pop soon, but LLMs as coding assistants won't go away.

AI models finally give us/the industry a reason to need more compute.

Replying rhetorically here, because this is a common point.

Wanting the Ai bubble pop and Ai to go away, can be mutually exclusive, which is a point also made, and I think people understand this. However, some might hope one achieves the other, which is less likely, highly.

In short: Bubbles, Tulips, we still have tulips!

I read it as an expression of frustration and some, the direct net effect on personal computer world costs, the spill over into corporate tulip mania, the overall net negative effect on life. It's an expression of a need to see life go back to some kind of normal.

It's been a tough ride since 2020. I don't blame people. The cost of compute is wrong on the growing list of many things wrong, and is part of the basket of what feels like hyper-inflation (as good as in terms of quality of life) every which way. People need stability, inflation, run away inflation kills so much dead fast.

 

[Omega Mac]

Apple stock about to explode like sandisk?

That's my thinking. (Declaration: No skin in the game).

When they start pumping Ai compute X times faster in the standups in Jobs Theatre, watch the markets reactions.

If they are seen to have or actually launch the iMac of Ai, then it has to be yes.

 

[Piggie]

Indeed, there is never going to be a POP it all stops.

There will be little bubbles growing rising to the surface and popping all the time.
Paths that are found to go nowhere.
Designs that prove to be the winning design, whilst others get left behind.

We can look at this like evolution if you like, the best things survive to the next generation.
It will of course slow, some may well say it's slowed already in certain areas.

No one know for sure yet if throwing more compute at the current models will simply reach a point where no gains can be found.

And of course we have others working on the software (brain) as they think the current models can only go so far and will hit a plateau.

Bit like the human running speed record, the human design can only go so fast before the change the design.

One thing is for sure, whilst someone can show their thing can do something someone else's thing cannot do, then the money will keep flowing, as everything is to play for here.

Whether after, perhaps 100 years from now, humans are living the best lives of luxury they have always dreamed of, or we are all wiped out by our robotic masters.... Well, we're just going to have to find out 😀

 

[thejadedmonkey]

Does this story sound odd to anyone else? Apple hosted "a 'Business at the Park' event involving executives from major companies Ford, Disney, and Anthropic" and yet also "did not possess... staff focused on developer relations..."

How does Apple even create and put on an event like that without such staff? Do they want us to believe that this was completely created by consumer-facing staff only?

Make it make sense, The Information.

A few months ago there was a story about Apple laying off some of their sales people in the business unit, and instead relying on partners to carry the load. If they're at that level of reduction of B2B, it's no wonder they're completely unprepared for business sales, in general.

 

[andiwm2003]

I keep wondering when some company - a new start-up, perhaps? - comes out with a product/device that allows one to use dramatically less RAM while getting even better performance than we do now and costing dramatically less than we pay now. Yes, I’m sure keepers of the status quo will say that’s impossible. But almost nothing really is impossible. I mean, if the market is grossly distorted and isn’t providing us with what we need now, then history tells us something is likely to come along to change that situation. I’ll bet the answer is simpler than we might think.

I always wonder if it was possible to make a chip that has the entire model with the weights as hardwired connections on the chip. A brain has the "weights" between the nerve cells hardwired (to some extent). Of course you would have to make a new chip if the model gets changed.

 

[smaffei]

AI has its uses; but using it for damn everything is starting to really annoy me. It promotes laziness. I am glad I only have 10 or more years left of working before retiring and not having to deal with this BS.

Amen!

I'm 57 and have been an Apple developer for 38 years. Can't wait until 65 so that I can get out of this industry. It's nowhere near what it was even 10 years ago. There is very little caring about quality in software anymore. Ship it fast and deal with the quality aftermath later. And now, with AI acceleration, things are even faster; ideas become apps in days with limited or no real QA.

AI slop is real and becoming the norm. I'm done.

 

[smaffei]

Apple is still missing the point…

The real reason the mini sold was because it is (was) a great machine for the price point. It's the same for the MacBook Neo whose high demand supposedly took Apple management by surprise too.

Regardless what some people want to think, the economy is bad for a lot of people out there. Many can't afford really expensive computers but would love to have a Mac. The Mini and Neo helped those folks get an Apple computer.

 

[krspkbl]

We all know Apple was caught of guard by AI. They've only just managed make Siri more intelligent than a brick and it still sucks compared to ChatGPT/Claude/Gemini etc. Apple is stuck in 2023 while everyone keeps pumping out new models almost by the week.

Tim was good for some things but AI isn't one of them. Good thing he's gone.

Apple could make a lot of money if they invest more in AI. I shouldn't be saying that though as a consumer. Would be good for the shareholders though.

 

[chrono1081]

Yup and we were caught off guard by the massive price increase. From $599-$899 in a few months.

No? Everyone knew it was coming. Apple staved off the massive price increases that was hitting the PC market for a year and a half, it was only a matter of time.

 

[zgagato]

It's like when people were crazy about Pokemon go !

Yea, the exact same.

 

[Warped9]

Buy a machine that runs all your needs on the device or pay endless absurd subscription rates for cloud service while not being able to guarantee the security of your data. It’s a no-brainer.

 

[Piggie]

I always wonder if it was possible to make a chip that has the entire model with the weights as hardwired connections on the chip. A brain has the "weights" between the nerve cells hardwired (to some extent). Of course you would have to make a new chip if the model gets changed.

I think that's either just been done or is about to.
Can't recall the company name.

And it may be been more "flashed" onto the chip?

It's certainly something that I sure many are or will be working on.
The thing is, as you can understand, you don't lock in a design in any industry when things are moving super fast.

It would be like locking down the iPhone with the iPhone 3, 4, or 5.

Now we are the iphone 17, one can easily argue that improvements are now so minimal, and almost playing the tiny adjustments, one could lock it down for 5 years and most would never care.

So, yes I 100% expect what you want to see will happen, and show be great, but I don't feel it will be a super popular concept until improvements slow considerably.

Found the article:

Google Frozen chip: Gemini baked into the silicon

A report says the Google Frozen chip would hardwire Gemini into silicon for up to 10x efficiency by 2028. What it means for AI.

thenextweb.com

Also:

Taalas | The model is The Computer

We are creating a platform that makes it easy to turn any deep learning model into custom silicon quickly.

taalas.com

Or course this is all very much "in the future" and for the high end.
But logically and eventually, given enough years passing one must presume that something like this for a home/end-user will be the goal for many companies.

 

[crsh1976]

how, exactly, was Apple "caught off guard" by releasing these models earlier than originally planned?

Desktops were never a bigger priority than notebooks, until the local LLM trend hit and demand for desktops exploded.

Some folks at Apple were caught off guard because they never saw it coming, as per Apple consistently having delayed reactions to anything AI. It caught their attention because there's more money to be made while the demand is still there (which may or may not be temporary).

 

[wanha]

Desktops were never a bigger priority than notebooks, until the local LLM trend hit and demand for desktops exploded.

Some folks at Apple were caught off guard because they never saw it coming, as per Apple consistently having delayed reactions to anything AI. It caught their attention because there's more money to be made while the demand is still there (which may or may not be temporary).

I agree with your overall point of Apple being late on AI in general, but as it pertains to this particular article and its specific context, saying "caught off guard" implies that Apple were caught unprepared.

And yet, being able to rise to meet the market by offering these machines ahead of their schedule implies the opposite of being unprepared.

 

[PBG4 Dude]

I was going to bring up arcade machines and home consoles, but figured hardly anyone would know about arcade machines.

In terms of home AI, we're approaching the ColecoVision era 🙂.

Core memory unlocked, lol. I used to deliver the morning paper to a developer on the Coleco ADAM team.

 

[crsh1976]

I agree with your overall point of Apple being late on AI in general, but as it pertains to this particular article and its specific context, saying "caught off guard" implies that Apple were caught unprepared.

And yet, being able to rise to meet the market by offering these machines ahead of their schedule implies the opposite of being unprepared.

Agreed, tho desktop sales were always steady but much lower volumes than notebooks - unprepared is right however, demand exploded for desktops and they couldn't right away as memory/storage are/were prioritized to sustain notebook sales in the first place.

 

[MiamiFL]

View attachment 2657042
Pop. Please pop

i want this bubble to burst so bad

When will this lunacy end.

Can’t wait for this psychopath-controlled bubble to burst.

There is no AI bubble to burst, at least not in the way doom and gloomers are claiming. There is hyper enthusiasm but the fakers will get shaken out very quickly. An actual bubble that bursts isn't possible anymore with all the market makers that exist now to expose fakers in the market very rapidly.

As Warren Buffet said, in the short term, the market acts as a voting machine, but in the long term, it acts as a weighing machine that measures a business's true value.

With market makers, that weighing of a business's true value now happens quickly and regularly.

 

[Biro]

I always wonder if it was possible to make a chip that has the entire model with the weights as hardwired connections on the chip. A brain has the "weights" between the nerve cells hardwired (to some extent). Of course you would have to make a new chip if the model gets changed.

That’s what I’m talking about. Thinking outside of the box. People are posting how the things I’m proposing are impossible with the current technology. We already know that. I’m talking about new technology that changes the landscape. These things tend to come out of nowhere.

 

[nt5672]

Apple, caught off guard by AI? Never saw that before 😂

In all seriousness, I am also ready for the lunacy to end, @Jakol!

which one,

The lunacy of Apple having no idea what is going on in the industry

or

The industry going all in for AI like they did for tulips. (https://en.wikipedia.org/wiki/Tulip_mania)

 

[marco114]

they wont even let me lease one of the 256GB models, only the 96GB. Lame.