Youâ(TM)re forgetting the Touch Bar. You canâ(TM)t accuse Apple of not trying to innovate, but it is a disaster. I canâ(TM)t type on my MBP without constantly triggering it. Thank goodness I use an external keyboard 90% of the time.

And given that most of my usage is with it on a stand as a secondary screen, Iâ(TM)d never find the use for it being a touch screen.

If you are sitting down and working, a mouse is better than either. (Though for real productivity, learn those keyboard shortcuts!)

If you are moving around, or doing standing demos, you want a 2-in-1. (360 hinge or detachable).
What a pity they can't just offer macOS on the ipad Air/Pro. But thats like asking for ports on a macbook air.

No. I meant what I wrote.

I imagine a touch screen would be most useful when you don't have convenient access to the pointer. For example, holding a laptop up to discuss something with a coworker. It would allow for the coworker to move windows around, switch tabs, etc, without having to physically hand over the laptop.

But if you are sitting there with your hands on the keyboard and thumb on the trackpad -- a touchscreen has no real use. But that is not the only scenario in which such devices are used.

Person A: Go to that link...

Person B: This link?

Person A: No, go down some more.... Now up.... No, not that one....

Person B: This link?

Person A: Great, now open it in a new tab... I said a new tab!!! Better go back, we need that previous page. No, don't close the window....

Am I the only one who can see this sort of interaction taking place? There are times when it could be very useful for "Person A" to just touch the screen.

I can only speak to myself. I don't use a laptop too frequently, desktop is primary, but I'd say a touch screen is/would be useful about 30-50% of the time while on it. I often find myself wanting to just tap on things, with Windows and MacOS. It's usually limited to casual usage.

While I wouldn't assume touch is the primary desired method of interaction, I know there can be times where it would be easier, or more intuitive, than using the touchpad. Further, with Apple integrating iPad apps and widgets into MacOS it seems at least somewhat more likely to be useful/desirable. From a function standpoint, it's a feature that cannot easily be handled via aftermarket. As such, it's generally better to have the option for those that need/want it. The only real reason to keep it out is cost.

I personally never cared for it, it was always felt awkward to me unless it was a 2-in-1 folded over and I was using it like a tablet. I was just wondering how big the demand was, I don't think it's as big a demand as Apple management might think.

BTW, typing this on an LG Gram 17 (from a few years back). Love it. Super light (less than 3 lbs), tiny bezels, packs a 17" screen in a laptop only marginally bigger than most 15.6"ers. And with a numeric keypad.

Indeed. Apple got it when they put it on a Unix basis.

Apparently, this was not tested. And this is a "you are fucked" level bug. Well, I guess reinstalling Windows may (or may not) fix this. Welcome to amateur-hour.

At the same time, Linux recovery (for which you have multiple options) continues to work just fine.p

512 Gb of storage is a sick joke in 2025. Single programs sometimes require more than that. The most expensive possible configuration they announced which has 512gb of storage costs about 3k. For 3k you can build a desktop server computer with more than 512gb of RAM, with a 1-4tb ssd depending on what kind of gpu and quality of ssd you want. Its not a terribly useful configuration for most use cases because you are spending almost all the money on an obsolete server platform and a ton of ram, but the fact you can do it is insane.

No worries. "AI" is such a vague word, and person you replied to mixed AI and LLMs which are the same, but also not the same. There wasn't a great correct answer :P

They asked about LLMs.

A Mac is, in fact, the best LLM inference machine you can buy realistically.

You will need to spend ~$30k on GPUs in order to run models you can run on your Mac.

You can also run those models on something like an Ryzen AI Max 395+, or one of those Broadwell mini computers, but the M4 Max is twice as fast, and the M3 Ultra even faster.

M3 Ultra comes with the additional benefit of being able to be loaded with 4x the VRAM of the M4 Max (already 128GB), making it worth 6-figures worth of GPUs.

Technically, certain stacks can use tensor parallelism (vLLM, LangChain) which means the performance scales with the GPUs, not just the VRAM- but nothing int he open source ecosystem really supports those.

Means realistically, Macs are the best thing you can buy for local LLM inference. Nothing else really competes.

I know that the main Apple motivation is to massage the investors, but is people really using their Macs to run local LLMs? (Genuine question, I have no idea). Another question: are software that have real AI capabilities using those NPUs or just exchanging data with company servers?

Given ethics and mandates of journalism, I would argue that signing such an agreement makes the information no longer news. So they have to choose, do news or do PR.

For all software, including embedded, the full source code shall be but into the national library of the country it is sold in - and it must be shown that it actually compile to the sold software.

People like will complain that Microsoft only supports hardware for a minimum of 9 years and complain about e-waste, while boasting they use Apple who will only support their computers for 7 years....