"This works with any Intel Mac app" [quoted from the article, regarding Rosetta 2]
Are you sure? Does that include Parallels running x86-64 Windows? It's quite telling that they mentioned Rosetta and virtualization, yet made no mention of this, which could alleviate concerns on many pro users' minds (myself included).
I doubt they meant that. But as Apple has said, only 2% of Macs coming in for service had Windows installed in Bootcamp. How many are using Parallels or other virtualization software with Windows, I don’t know, but it’s not a lot. I have it too, but I haven’t run Windows for more than a year. I still do Run Linux occasionally though. So likely, from what I hear, that’s more important.
i doubt I’d too many pro users use Windows on their Mac these days. It’s mostly used by gamers.
I have zero games on my Windows installation under Parallels. I do have EDA software (electronics simulation, schematic capture, PCB routing, FPGAs, etc.), test & measurement software to interface with electronics T&M gear, MCAD software, software development apps (Visual Studio, the real one not the toy Code version, plus various embedded software tools), etc.
Another group of people will have in-house apps that are Windows only.
Maybe in your line of work pro users don't need Windows software. It doesn't mean no one else does.
And those like you consist of what, 0.5% of Apple’s user base?
Facepalm. If it were that tiny - as Apple has only 6-8% of the computer sales in any given year - Parallels wouldn't be commercially viable in the first place. It is precisely because there are LOTS of macOS users who need Windows software for work. It looks like Google and Parallels are teaming up to promote Chromebooks and Chromeboxes to replace MacBooks and iMacs for this use case. Parallels clearly prefers macOS users continuing to use their products on ChromeOS to them switching to Windows and there is a reason for that.
"This works with any Intel Mac app" [quoted from the article, regarding Rosetta 2]
Are you sure? Does that include Parallels running x86-64 Windows? It's quite telling that they mentioned Rosetta and virtualization, yet made no mention of this, which could alleviate concerns on many pro users' minds (myself included).
By the time Apple switches to ARM, Windows would have a better version on ARM as well.
But it will be compatible with the Qualcomm/MediaTek/Samsung ARM and not Apple A14 ARM which has a different architecture and instruction set. If Apple ARM and Qualcomm ARM were compatible, people would have installed iOS on Samsung Galaxy hardware - for example - or vice versa ages ago just to say that they could. They don't because Android won't run on Apple Ax and iOS won't run on Samsung/Exynos chips because the arch and instruction sets are too different. The same will apply to Windows on ARM.
Glad to see we’re heading towards the 22nd century with a better architecture and software platform. Now if only the Windows platform would just die already.
If any platform dies it will not be the one that has 80% market share and like 95% compatibility with custom professional/enterprise applications. Also, within 3 years Microsoft, Intel and AMD will figure out a solution for emulating x86 and x86-64 on mainstream ARM (meaning Qualcomm, MediaTek and Samsung). Those are about to have a massive leap forward in performance thanks to the Cortex X1 super core that ARM Holdings and Samsung co-developed. In addition, the first batch of 5 nm (non-Apple) midrange ARM CPUs are already available in devices being sold in China. The first flagship devices with them are going to be available in the United States in February. The second generation of those chips available in 2022 will likely be when x86 and x86-84 emulation will be practical. At that time, Windows users will be able to pick Qualcomm (or Samsung), AMD or Intel devices with roughly similar performance, and ChromeOS ones also.
Speaking of Intel, they aren't going to just stand pat. They are having TSMC manufacture their new 7nm design rather than wait on their own foundries. Right about the time that Apple ends support for Intel-based Macs, they should be on 5nm. Of course, Apple will be on 3nm for their Ax chips by that time, but the point is that when Intel reaches 5nm - whether their foundries or TSMCs' foundries actually manufactures the chips it will happen soon - their chips will have high performance/low power designs too. Even more so for AMD, who like Qualcomm and Samsung will release 5nm chips in 2021 (though a 5nm AMD chip is akin go a 7nm Intel chip in terms of performance, but that is a whole other conversation).
Glad to see we’re heading towards the 22nd century with a better architecture and software platform. Now if only the Windows platform would just die already.
If any platform dies it will not be the one that has 80% market share and like 95% compatibility with custom professional/enterprise applications. Also, within 3 years Microsoft, Intel and AMD will figure out a solution for emulating x86 and x86-64 on mainstream ARM (meaning Qualcomm, MediaTek and Samsung). Those are about to have a massive leap forward in performance thanks to the Cortex X1 super core that ARM Holdings and Samsung co-developed. In addition, the first batch of 5 nm (non-Apple) midrange ARM CPUs are already available in devices being sold in China. The first flagship devices with them are going to be available in the United States in February. The second generation of those chips available in 2022 will likely be when x86 and x86-84 emulation will be practical. At that time, Windows users will be able to pick Qualcomm (or Samsung), AMD or Intel devices with roughly similar performance, and ChromeOS ones also.
Speaking of Intel, they aren't going to just stand pat. They are having TSMC manufacture their new 7nm design rather than wait on their own foundries. Right about the time that Apple ends support for Intel-based Macs, they should be on 5nm. Of course, Apple will be on 3nm for their Ax chips by that time, but the point is that when Intel reaches 5nm - whether their foundries or TSMCs' foundries actually manufactures the chips it will happen soon - their chips will have high performance/low power designs too. Even more so for AMD, who like Qualcomm and Samsung will release 5nm chips in 2021 (though a 5nm AMD chip is akin go a 7nm Intel chip in terms of performance, but that is a whole other conversation).
Good to see a recognition that Intel isn’t going to shrivel up and die because Apple is going their own way on their own machines. Intel still has a huge market that they serve outside of Apple and outside of PCs. I tend to think that software is the primary driver for processor selection, and in most cases software is the predominant cost driver, at least for systems.
"This works with any Intel Mac app" [quoted from the article, regarding Rosetta 2]
Are you sure? Does that include Parallels running x86-64 Windows? It's quite telling that they mentioned Rosetta and virtualization, yet made no mention of this, which could alleviate concerns on many pro users' minds (myself included).
Compare that to other pro gear. I work with electronics design, where you can get upgrades for decades-old test equipment from the likes of Keysight, Fluke or Tektronix. An HP 3458A DMM, the gold standard in high-precision metrology, is a design from 1989 (IIRC) which holds its value quite well, and is still sold today with minimal, user-facing changes only. The lens mounts for DSLR cameras are the same for decades, you can use a good lens from the previous century on a current Canon or Nikon camera. I know computer technology is faster paced than this, but still, the timeframes in the pro market are quite different from the consumer market.
What a silly comparison. Show me any decades old Windows PC. I am using a decade old Mac Pro as my main desktop still
I went back through the Keynote and at the 1:40:11 mark, Docker (docker.com) was shown running Linux. At the 1:41:58 mark Parallels was shown running Debian. Craig said all macOS Big Sur demoes were run on an AS Mac so I assume it's either the AS Mac mini or another development AS Mac. Parallels has made some big changes in ver 15 but I run VMWare Fusion so haven't looked at Parallels for a long time. Anyway, at this point in the keynote they were talking about Rosetta 2 so I assume they simply installed Parallels ver 15 and it converted it to run on Apple Silicon.
Parallels on Intel doesn't need to emulate x86 since an x86 chip. Parallels on ARM would need to be what Parallels was back in the PowerPC days - emulating x86 instructions on a non-native x86 chip.
That's what they demoed during WWDC, although I agree - they did a poor job of really clarifying that.
You don't know that. They're a much bigger company now and with their announcement that they have Intel Macs in the pipeline I think you can assume you have meaningful support for at least a few years beyond that. They're not going to abandon everyone. Just give it a minute. They're not going to announce every single detail right now.
PowerPC support was a good five years - and they were in a much more precarious financial situation back then. I think with the whole Mac Pro thing Apple is once again paying attention to and valuing the importance of the Mac so I'm not as concerned. I still don't blame others for being skeptical given Apple's past foibles, especially when we are talking $10K plus machines. Skepticism is simply prudent.
I suspect AS going to be so significant a sea change that most of these concerns will take care of themselves - if not, why expose themselves to all the concerns outlined in this thread? If they weren't confident that the benefits of AS were so compelling that if you really were a pro user the benefits would more than make up for the cost of changing then this would be their biggest folly ever.
Apple has done some dumb things in the past, but even they couldn't be that stupid. Change can be unsettling, but it can also be exciting. None of our current machines will quit working when the AS Macs come out, so let's see where they start before declaring it's the end of the world (again)
To see how many real Mac users Apple will lose, it's necessary to find reports of the percentage of users running VMware, Parallels or VirtualBox. Note that I said "reports" there, not "numbers taken out of one's rear end". Without that figure, the discussion cannot proceed.
If you are using Parallels today why couldn't you use it tomorrow? It's clear they have x86 emulation - just like they did with PowerPC; they have done this before.
Of course with software emulation of hardware, performance is always an issue. Fun fact - Intel x86 instructions don't execute natively within Intel CPUs. It's not the same as virtualization but it is roughly analogous. Note Parallels stated they were working closely with Apple - I suspect that's an understatement. If there were some hardware optimizations that could dramatically speed up virtualization, Apple owns the silicon. Theoretically could build in extra support to support x86 emulation and make it not fully redundant on software. I'm sure there are interesting patent issues but Apple has pretty good attorneys and has also been buying lots of companies that have their own patents. Patent horse trading is a thing. And Apple still has decent relationships with Intel or the whole cellular modem thing wouldn't have happened, and I guarantee you Apple Silicon was NOT a surprise to Intel (just like the move from PowerPC to Intel wasn't a surprise to Motorola/IBM).
I think the ability to customize Apple silicon is one of its greatest strengths - Apple owns and can customize every aspect of it. We lost display mirroring on the iMacs because Intel didn't support the neded resolutions and Apple had to hack a custom display controller.
No more need to hack. They can design exactly what they need, exactly where and when they need it.
Yes, there is going to be some short term instability and questions. Long term? There's probably stuff we aren't even thinking about because up to now we have all (including Apple) been restricted by least common denominator mass market parts as the foundations for all of our computers.
Not any more. The thought of what a company like Apple, who has no problem obsoleting their own products with a new product (iPod Mini to iPod Nano or all iPods with the iPhone, etc.) having essentially a blank slate to re-imagine general purpose computing hardware *and* software? Gives me goosebumps.
Any time you obsolete a product, you take a huge chunk of people from the pool of potential buyers (i.e. the demand) while the supply is kept identical (seeing as nobody invented a time machine that allows you to go back in time and not purchase the Mac Pro that you already did). There is no other possible outcome than the price of the goods being reduced under this scenario.
Not unless your new product attracts more customers than you could have gotten with your old one.
Pro apps like Avid And Adobe don't use the Apps store. But It's good to see Adobe a App running in the WWDC demo on Arm.
Indeed! I think Apple is still smarting a bit from Adobe's Carbon/32bit fiasco so I was rather amused to see "native" creative suite at the keynote. All kinds of subtext there if you have any real history with Adobe!
cloudguy said: Speaking of Intel, they aren't going to just stand pat.
lol - I wouldn't brag that Intel is using TSMC. That is not a point of strength.
Intel has been standing pat for the past 7 years. It's why we are here. They failed, miserably, at meeting their performance roadmaps - in compute performance, power and thermal envelope. If Intel had been doing anything but "standing pat" I doubt we would be talking about an Apple Silicon transition *now*. It would have probably still happened, but probably a few years from now and maybe just a few Macs. Not a wholesale replacement in two years. That is a HUGE vote of no confidence, PERIOD in Intel being able to deliver.
Amazon is chomping at their Xeon business with their own Arm ventures - Intel better start doing more than releasing power points of roadmaps.
It was frustrating to rely on Intel's processor schedule to make any plan of new MAC releases. So, Apple Silicon based MAC was needed for a long time to control it's destiny.
So do expect some complaints, and also expect some bargain Intel-based Mac Pro machines to turn up on eBay. However, it's not that anyone need ditch their current Intel Mac, nor should anyone should put off buying one if they need it now.
Something I've never understood about some users. Your current machine is running perfectly fine, it's fast and it does what you want it to very well. Now something new and different comes along and somehow, someway , the machine you are using becomes an obsolete piece of crap not worth keeping. And you blame Apple for bringing out a new technology before you are damn good and ready for it. You rage at Apple for making your perfectly fine machine 'useless'.
When you invest a substantial amount of money in some pro gear, you hardly do so with the expectation to use it for a couple of years and then discard it. In fact you sell it for a good fraction of what you paid for it. That's part of the economic calculation of buying a piece of pro gear.
Now suddenly your pro gear uses a fundamentally incompatible architecture, which will be supported for "some (unstated amount of) years". There's no guarantee developers will continue performing software maintenance for the Intel port, or even Apple itself, for that matter. Now your expensive pro gear may not last as long as you initially planned, and by the time you sell it, it will probably be worthless. I mean really, if you paid upwards of $10,000 on a Mac Pro recently (quite easy with CPU, RAM, storage and GPU upgrades), who's going to pay more than, say, $3,000 or $4,000 for it in three years, knowing the fate of Intel hardware?
Compare that to other pro gear. I work with electronics design, where you can get upgrades for decades-old test equipment from the likes of Keysight, Fluke or Tektronix. An HP 3458A DMM, the gold standard in high-precision metrology, is a design from 1989 (IIRC) which holds its value quite well, and is still sold today with minimal, user-facing changes only. The lens mounts for DSLR cameras are the same for decades, you can use a good lens from the previous century on a current Canon or Nikon camera. I know computer technology is faster paced than this, but still, the timeframes in the pro market are quite different from the consumer market.
If Apple really cared about its pro users, they should have stated Mac Pros will be supported by macOS and pro apps for, at the very least, 5 years, and for them to keep a modicum of resale value, 10 years. They could go even further by requiring fat Intel/ARM builds in the Mac App Store for a similar amount of time, but macOS and pro app support for 5-10 years is the bare minimum.
There are pro markets where innovations are slow or not needed because the underlying requirements not changing. That's the pro market you're referring to.
Then there is the pro market, say for space exploration, which takes 20-30 years from inception to launch where locking in the technology is required early in order to harden the design. Keeping up with technological advances is impossible in this case.
Then there are the pros in the movie business who will always move to the next big thing because the advantages of bigger and faster is far outweighed by cost of abandoning mere two year old hardware.
I went back through the Keynote and at the 1:40:11 mark, Docker (docker.com) was shown running Linux. At the 1:41:58 mark Parallels was shown running Debian. Craig said all macOS Big Sur demoes were run on an AS Mac so I assume it's either the AS Mac mini or another development AS Mac. Parallels has made some big changes in ver 15 but I run VMWare Fusion so haven't looked at Parallels for a long time. Anyway, at this point in the keynote they were talking about Rosetta 2 so I assume they simply installed Parallels ver 15 and it converted it to run on Apple Silicon.
Parallels on Intel doesn't need to emulate x86 since an x86 chip. Parallels on ARM would need to be what Parallels was back in the PowerPC days - emulating x86 instructions on a non-native x86 chip.
That's what they demoed during WWDC, although I agree - they did a poor job of really clarifying that.
Parallels didn't exist before Intel Macs came along. Connectix VirtualPC was the only commercial x86 emulator, and it was incredibly slow. MS eventually bought it and shut it down.
No they didn't demo x86 on ASi, they demoed Linux, which was quite likely to have have been an ARM build.
It was frustrating to rely on Intel's processor schedule to make any plan of new MAC releases. So, Apple Silicon based MAC was needed for a long time to control it's destiny.
Nothing to do with Intel. After about 2011 (after Jobs died, what a coincidence) Mac updates were getting less and less frequent. The Mac Pro went 6 years without a real update, the iMacs went almost 3 years without a major update, the Mac Mini 4 years, etc. Intel released CPUs in those timeframes, but Apple was just ignoring the Mac. There is more than just a CPU in the Macs too, GPUs, storage, RAM etc could have been upgraded without a processor bump.
"This works with any Intel Mac app" [quoted from the article, regarding Rosetta 2]
Are you sure? Does that include Parallels running x86-64 Windows? It's quite telling that they mentioned Rosetta and virtualization, yet made no mention of this, which could alleviate concerns on many pro users' minds (myself included).
By the time Apple switches to ARM, Windows would have a better version on ARM as well.
But it will be compatible with the Qualcomm/MediaTek/Samsung ARM and not Apple A14 ARM which has a different architecture and instruction set. If Apple ARM and Qualcomm ARM were compatible, people would have installed iOS on Samsung Galaxy hardware - for example - or vice versa ages ago just to say that they could. They don't because Android won't run on Apple Ax and iOS won't run on Samsung/Exynos chips because the arch and instruction sets are too different. The same will apply to Windows on ARM.
This is completely wrong.
Okay, let's go back over the past five years and cover the important details that many people are still missing.
Apple has the most expensive ARM license available, which ironically gives them nothing. All they get is a book detailing the instruction set, and a compatibility test suite to make sure that their in-house designs are implementing the instruction set correctly. Apple is free to add new instructions to optimise it for their own requirements, but every design they produce runs through this compatibility suite.
They don't because Android won't run on Apple Ax and iOS won't run on Samsung/Exynos chips because the arch and instruction sets are too different.
If iOS won't run on Android then it's because Apple it uses instructions that are not in the ARM spec. However, Android cannot use an extended instruction set because it doesn't know what ARM chips it will be running on, and same goes for Windows ARM.
If what you're saying was correct then Apple would've been unable to demonstrate Linux and Docker running on ASi at the last WWDC. There is no technical reason why Windows ARM can't run on ASi as long as Apple keeps passing the compatibility tests.
But the operating system is only half the story. Apps will need to be recompiled, and for folk running obsolete software that's not been written to Microsoft's architecture-neutral guidelines, then this day was bound to come.
What Apple has done will be a massive boost to ARM across the board, a bit like Apple adopting USB.
It was frustrating to rely on Intel's processor schedule to make any plan of new MAC releases. So, Apple Silicon based MAC was needed for a long time to control it's destiny.
Nothing to do with Intel. After about 2011 (after Jobs died, what a coincidence) Mac updates were getting less and less frequent. The Mac Pro went 6 years without a real update, the iMacs went almost 3 years without a major update, the Mac Mini 4 years, etc. Intel released CPUs in those timeframes, but Apple was just ignoring the Mac. There is more than just a CPU in the Macs too, GPUs, storage, RAM etc could have been upgraded without a processor bump.
When you ask the general public why they're upgrading their machine, the most popular answer will be 'Because I needed a faster machine.' Most people will struggle through with decreasing disk space until they can get a faster machine. Apple would find it quite difficult to sell the notion of upgrades to the user base simply by increasing RAM and changing the GPU. Folk don't care about that as much as they care about increased performance.
But the bigger problem is that Apple has plotted a course that will need lots of power that runs very cool. It's unlikely that Intel can do both.
It was frustrating to rely on Intel's processor schedule to make any plan of new MAC releases. So, Apple Silicon based MAC was needed for a long time to control it's destiny.
Nothing to do with Intel. After about 2011 (after Jobs died, what a coincidence) Mac updates were getting less and less frequent. The Mac Pro went 6 years without a real update, the iMacs went almost 3 years without a major update, the Mac Mini 4 years, etc. Intel released CPUs in those timeframes, but Apple was just ignoring the Mac. There is more than just a CPU in the Macs too, GPUs, storage, RAM etc could have been upgraded without a processor bump.
When you ask the general public why they're upgrading their machine, the most popular answer will be 'Because I needed a faster machine.' Most people will struggle through with decreasing disk space until they can get a faster machine. Apple would find it quite difficult to sell the notion of upgrades to the user base simply by increasing RAM and changing the GPU. Folk don't care about that as much as they care about increased performance.
But the bigger problem is that Apple has plotted a course that will need lots of power that runs very cool. It's unlikely that Intel can do both.
Of course, but that's usually after 5 or 6 years, in which case almost any computer they get will be faster. But people are rightly unhappy paying the same price for a Mac that's got 4 year old old tech in as one that has the latest, because Apple never drops their prices as things age. That's no problem when things are updated regularly, but not with a 4 year gap. And in any case, Intel didn't go 4 years between CPU updates, Apple could easily update the CPU each time Intel does and it would always be quicker - albeit not much.
They have, but half of that objective is entirely form over function. My iMac can't keep its i9 cool enough to run much above the base clock due to insufficient cooling, because it's too thin. The GPU throttles because it gets too hot. The CPU heat issue would be resolved by ASi - the GPU issue won't - but both would be resolved by making the machine slightly thicker. A desktop does not need to be thin, that's a Johnny Ive obsession that I hope now is going to be relaxed.
Anecdotal, but the general populace I’m in contact with upgrade their machines only when they stop working.
When things start going wonky and unreliable, or they need compatibility with some piece of hardware or software no longer supported by their machine. THAT is when machines are upgraded.
Until then, the computer continues to do everything it has always done, at pretty much exactly the same speed.
This, incidentally, includes many of the “professionals” I work with — the tracking Mac Pro in the studio space I share is over ten years old.
One of the mixing iMacs is a 2014 27”, running High Sierra.
Unless you need compatibility with the latest hit plug-ins for that super popular sound du jour, mixes made on the newer machines don’t any better, and audio really doesn’t run any faster on a new machine.
My personal machine is used as a stage computer as well (or will be again at some point when this whole mess is over), so I keep it more or less current for MainStage and compatibility with my iPhone and iPads. The last one lasted about six years before it started getting unreliable. I’m four years into this one, and it’s still kicking strong.
Comments
Speaking of Intel, they aren't going to just stand pat. They are having TSMC manufacture their new 7nm design rather than wait on their own foundries. Right about the time that Apple ends support for Intel-based Macs, they should be on 5nm. Of course, Apple will be on 3nm for their Ax chips by that time, but the point is that when Intel reaches 5nm - whether their foundries or TSMCs' foundries actually manufactures the chips it will happen soon - their chips will have high performance/low power designs too. Even more so for AMD, who like Qualcomm and Samsung will release 5nm chips in 2021 (though a 5nm AMD chip is akin go a 7nm Intel chip in terms of performance, but that is a whole other conversation).
What a silly comparison. Show me any decades old Windows PC. I am using a decade old Mac Pro as my main desktop still
Parallels on Intel doesn't need to emulate x86 since an x86 chip. Parallels on ARM would need to be what Parallels was back in the PowerPC days - emulating x86 instructions on a non-native x86 chip.
That's what they demoed during WWDC, although I agree - they did a poor job of really clarifying that.
I suspect AS going to be so significant a sea change that most of these concerns will take care of themselves - if not, why expose themselves to all the concerns outlined in this thread? If they weren't confident that the benefits of AS were so compelling that if you really were a pro user the benefits would more than make up for the cost of changing then this would be their biggest folly ever.
Apple has done some dumb things in the past, but even they couldn't be that stupid. Change can be unsettling, but it can also be exciting. None of our current machines will quit working when the AS Macs come out, so let's see where they start before declaring it's the end of the world (again)
If you are using Parallels today why couldn't you use it tomorrow? It's clear they have x86 emulation - just like they did with PowerPC; they have done this before.
Of course with software emulation of hardware, performance is always an issue. Fun fact - Intel x86 instructions don't execute natively within Intel CPUs. It's not the same as virtualization but it is roughly analogous. Note Parallels stated they were working closely with Apple - I suspect that's an understatement. If there were some hardware optimizations that could dramatically speed up virtualization, Apple owns the silicon. Theoretically could build in extra support to support x86 emulation and make it not fully redundant on software. I'm sure there are interesting patent issues but Apple has pretty good attorneys and has also been buying lots of companies that have their own patents. Patent horse trading is a thing. And Apple still has decent relationships with Intel or the whole cellular modem thing wouldn't have happened, and I guarantee you Apple Silicon was NOT a surprise to Intel (just like the move from PowerPC to Intel wasn't a surprise to Motorola/IBM).
I think the ability to customize Apple silicon is one of its greatest strengths - Apple owns and can customize every aspect of it. We lost display mirroring on the iMacs because Intel didn't support the neded resolutions and Apple had to hack a custom display controller.
No more need to hack. They can design exactly what they need, exactly where and when they need it.
Yes, there is going to be some short term instability and questions. Long term? There's probably stuff we aren't even thinking about because up to now we have all (including Apple) been restricted by least common denominator mass market parts as the foundations for all of our computers.
Not any more. The thought of what a company like Apple, who has no problem obsoleting their own products with a new product (iPod Mini to iPod Nano or all iPods with the iPhone, etc.) having essentially a blank slate to re-imagine general purpose computing hardware *and* software? Gives me goosebumps.
Not unless your new product attracts more customers than you could have gotten with your old one.
iPod to iPhone isn't just a theory
Then there is the pro market, say for space exploration, which takes 20-30 years from inception to launch where locking in the technology is required early in order to harden the design. Keeping up with technological advances is impossible in this case.
Then there are the pros in the movie business who will always move to the next big thing because the advantages of bigger and faster is far outweighed by cost of abandoning mere two year old hardware.
No they didn't demo x86 on ASi, they demoed Linux, which was quite likely to have have been an ARM build.
Nothing to do with Intel. After about 2011 (after Jobs died, what a coincidence) Mac updates were getting less and less frequent. The Mac Pro went 6 years without a real update, the iMacs went almost 3 years without a major update, the Mac Mini 4 years, etc. Intel released CPUs in those timeframes, but Apple was just ignoring the Mac. There is more than just a CPU in the Macs too, GPUs, storage, RAM etc could have been upgraded without a processor bump.
Okay, let's go back over the past five years and cover the important details that many people are still missing.
Apple has the most expensive ARM license available, which ironically gives them nothing. All they get is a book detailing the instruction set, and a compatibility test suite to make sure that their in-house designs are implementing the instruction set correctly. Apple is free to add new instructions to optimise it for their own requirements, but every design they produce runs through this compatibility suite.
That's not actually true.
https://www.androidauthority.com/android-on-iphone-1089705/
If iOS won't run on Android then it's because Apple it uses instructions that are not in the ARM spec. However, Android cannot use an extended instruction set because it doesn't know what ARM chips it will be running on, and same goes for Windows ARM.
If what you're saying was correct then Apple would've been unable to demonstrate Linux and Docker running on ASi at the last WWDC. There is no technical reason why Windows ARM can't run on ASi as long as Apple keeps passing the compatibility tests.
But the operating system is only half the story. Apps will need to be recompiled, and for folk running obsolete software that's not been written to Microsoft's architecture-neutral guidelines, then this day was bound to come.
What Apple has done will be a massive boost to ARM across the board, a bit like Apple adopting USB.
When you ask the general public why they're upgrading their machine, the most popular answer will be 'Because I needed a faster machine.' Most people will struggle through with decreasing disk space until they can get a faster machine. Apple would find it quite difficult to sell the notion of upgrades to the user base simply by increasing RAM and changing the GPU. Folk don't care about that as much as they care about increased performance.
But the bigger problem is that Apple has plotted a course that will need lots of power that runs very cool. It's unlikely that Intel can do both.
They have, but half of that objective is entirely form over function. My iMac can't keep its i9 cool enough to run much above the base clock due to insufficient cooling, because it's too thin. The GPU throttles because it gets too hot. The CPU heat issue would be resolved by ASi - the GPU issue won't - but both would be resolved by making the machine slightly thicker. A desktop does not need to be thin, that's a Johnny Ive obsession that I hope now is going to be relaxed.