It's a very similar situation with the iPhone. Apple makes a great phone in many respects, but they're one company and they can't meet the wide variety customers demand. That's reflected in their market share.
It's also too early to say that Apple Silicon is a threat to anyone. Apple Silicon in the iPhone and iPad is a low power & low performance chip compared to AMD/Intel. They have a huge challenge scaling that up to compete on desktop. Their first chips are going into their lower performance products. I'm sure it'll be decent in that role, but Apple likely NEEDS 5nm just so that it makes a respectable showing compared to AMD/Intel.
While true for the world for now, they already own half of the phones sold in US. In revenue/profit terms, it is even higher. They have already beaten everyone with a fabulous phone processor design. Imagine that coupled with fab advantage. While others can differentiate on other things but at some point, some differences will become hard to overlook.
They are at a dangerous position where they can suffocate others like this easily because no one else has comparable cash to outbid them for cutting edge stuff.
Such an advantage is not the end, though. Look at where Intel is, with all the R&D spending, fab, and anti-competition moves from a couple of years ago. Engineers move around, and it is difficult to keep a top team for decades. You always end up making mistakes that can be used by your competition.
They won't keep the fab advantage. The money TSMC gets from Apple will contribute to improving production, which will also benefit other TSMC customers. Just like Apple buying most of the world's supply of NAND chips a couple of years back did not kill the market.
ARM Macs may be different in the future, but I don't think the performance advantage in phones is that important. People chose mostly on features and ecosystem. Even Android geeks do not care that much that last generation's iPhone outperforms their current flagship. Android still sells. Now, more performance does give more room for some features, but this is a second-order effect. For now, you can find almost any iPhone feature in some Android phones, often with bigger batteries or other things like a headphone jack or a SD port. OnePlus' phones are slower, but it's not the end of the world.
A vertically integrated Apple will suffocate the others if they become dominant enough to utterly dominate their market, which I don't see happening as long as they focus on the high end.
But they aren't just focusing on the higher end. They have products at competitive prices now like iPhone SE or Watch SE/ older gen watch.
And unquestionably TSMC now has much more 7nm capacity for sale than they would have if Apple were vertically integrated and owned their own fabs.
Anandtech's Tiger Lake deep dive compares performance gains when running SPEC at a 15 watt TDP and a 28 watt TDP.
>At first glance, the Tiger Lake system performs quite well versus its predecessor, but that’s mostly only in the 28W mode. At 15W, the generational boost, while it is there, isn’t that significant.
https://www.anandtech.com/show/16084/intel-tiger-lake-review...
We're getting to the point where Intel is no longer able to deliver significant performance improvements inside a low TDP.
This bodes well for an architecture that has much better performance per watt.
>Here we present the 15W vs 28W configuration figures for the single-threaded workloads, which do see a jump in performance by going to the higher TDP configuration, meaning [Tiger Lake] is thermally constrained at 15W even in ST workloads.
Comparing it against Apple’s A13, things aren’t looking so rosy as the Intel CPU barely outmatches it even though it uses several times more power, which doesn’t bode well for Intel once Apple releases its “Apple Silicon” Macbooks.
https://www.anandtech.com/show/16084/intel-tiger-lake-review...
But why are we comparing with Intel chips, when it is AMD chips that have been more successful in delivering higher performance with lower TDPs?
>Against the x86 competition, Tiger Lake leaves AMD’s Zen2-based Renoir in the dust when it comes to single-threaded performance.
>[In multicore] AMD’s platform scales incredibly well in execution-bound workloads as it fully takes advantage of double the core count. In more memory-heavy workloads, the Zen2 cores here seem to be lacking sufficient resources and scale below the performance of Intel’s 4-core designs in some workloads.
We'll have to see what the testing figures look like after Zen3 and Apple Silicon ship.
I think you can run Linux, Windows, some of the BSDs, on them, can't you?
Linux works, but the latest models usually have bad driver support for a while. I'm using a 2-year-old MacBook Pro with Arch Linux and it works fine after some tweaking.
BSD is not well supported.
Apple is quickly turning from pushing he at a premium to pushing HW and services in a walled garden package.
Not sure that's been true for some time now. Wouldn't they be better described as "a premium manufacturer of mobile phones that run iOS with an app ecosystem they make 30% commission on, who also make legacy propitiatory laptop and desktop devices"?
(I agree with the rest of your comment though. Apple _don't want_ 99% of Android or Windows customers, and are more than happy for Huawei and Oppo and AliExpress to sell phones and laptops to people Apple dont care about. They _might_ care a bit about encouraging top of the range Samsung or Pixel phone customers to iPhones, maybe. And they at least used to try to woo the dev/webdev crowd to macOS machines, but less obviously these days, it seems to me.)
> If you get a PC, you get literally thousands of choices.
And none of them work reliably.
That is absurd. Last year I built a PC with off the shelf components that I spent about 30 minutes picking out, and all I had to do was stick the windows installer into the USB port to get everything set up perfectly out of the box. I have never had any reliability issues and I only reboot about once per month for updates. I used Apple products exclusively for a decade, but a contract role last year required me to use a Windows dev environment. I was holding my nose going in, but using WSL has been a joy and my MacBook is now collecting dust in the closet. In my experience most Mac purists that turn their nose up at commodity PCs haven't used one since MacBooks became the standard hip startup dev machine. Things have changed a lot in the last ten years.
And let's not pretend that Mac hardware has been perfectly reliable lately. In fact, of my hand rolled desktop PC and my 2017 MacBook Pro, the PC has had far fewer hardware reliability issues. Apple's declining hardware quality and increasingly unjustifiable retail prices have pretty much chased me off the platform after 10+ years of fanatical loyalty.
Sure companies fill them with shit, that’s not really a flaw in the underlying hardware/OS though. Those same people have less scope for that on OSX/Linux sure, but again that doesn’t mean the software/hardware is better.
Which has 0 relevance. GGP was asserting that Apple Is an existential threat to AMD. Given AMD is not an Apple supplier (of CPU anyway) and Apple will not sell their chips to the outside world, that makes sense on no axis.
Any sources for this? As far as I know, AMD has never designed CPU for mobiles.
It's not out of the question for them to consider such a move, but I think that the bigger issue here for AMD is that it means they will wind up a mode behind the edge, giving their main competitor- Intel- more time to catch up.
My experience is the opposite. In my team at work people have either recent Mac book pro or Thinkpad with Linux. And it seems like the ones on Macs have always all kind of issues (with zoom screen sharing, with openvpn/tunnelblick, with docker not working as expected, and lately, as it's hot in paris, cooling issues - I laughed as they were exchanging tips too cool their laptop with stuffs you can put in the freezer, because otherwise their computer would be so slow that some of our time sensitive tests would fail...)
If it makes sense for other companies to adopt these kinds of financial arrangements, they will eventually, and while it works out best for Apple here, at this particular point in time, this is just one point in time. It does benefit others in the long run as TSMC is able to refine their process fabbing chips for Apple and will be able to open new production lines in the future with the profits from manufacturing for Apple and offer those to their other customers.
AMD is still doing really well in the market, and their customers are different and their more direct competition is Intel and NVIDIA. Could they make better products on a 5nm process size? We assume they can and will, maybe that will happen a little later than they would like, or maybe they’re working behind the scenes and making their own deals with other manufacturers that we don’t know about yet. This isn’t the bad old days where Intel was completely dominant, competition actually is good here. Apple is ahead right now, but that’s a temporary situation much like anything else.
At this point in time, this is the situation, but no one here should expect this to remain the situation in perpetuity. From what I heard, prior to this, Apple and AMD were jointly buying up all of or almost all of TSMC’s 7nm fab capacity. AMD has risen from the the dirt, and then later again from the ashes and is presently making the best x86_64 chips you can buy today. I’m actually disappointed Apple didn’t try to shop around for the 2019 Mac Pro CPU because they could have gotten much better all-around parts buying the equivalents from AMD for less and with lower power consumption. Ah well, can’t win them all.
I‘m not sure that’s true. Afaik Intel parts still have a rather large advantage concerning idle power consumption.
It has nothing to do with 5nm, Apple, or not.
This is the kind of situation market dynamics excel at. The next few years of CPUs in all our devices are going to be fantastic.
It allows for some parts of chip parts to be smaller. Most transistors stay about the same, because you cannot make them much smaller, otherwise they will leak or will not be as fast as needed.
If you look for transition of the same company from lambda1 to smaller lambda2 you will see that gate or transistor density rises not as theoretical square of lambda1/lambda2, but as about lambda1/lambda2 or less.
It is also quite possible to screw things at smaller lambdas. For example, NVidia get their process right for same lambda only after several years, they were fighting steady state leaks, power issues and the like. And NVidia, just like AMD and Apple, was using TSMC.
So if you are pissed about AMD lagging behind a node, don't be. Being at forefront of node size does not mean instant success, lagging a node size step behind does not mean you will get much inferior product.
An other commenter theorised that Apple had simply outbid AMD for the production capacity.
In practice, I suspect these sorts of situation arise far more frequently than we think. One party overpays for a supply of an ingredient and then controls demand by leveraging it. In the first act, they convince people that they need things made with this ingredient. In the second, they bank on the cachet, the exclusivity, driving demand even higher.
But at the end of the day, someone is playing a chess match where in theory the customer could have had more options and now they don't. Or they could have had no options at all. We don't have a time machine so who can say?
Apple went farther in a number of situations by investing in suppliers in order to guarantee that they could produce an ingredient in the first place, either at all, or in a timely fashion. We are going to enable you to make a product, and then we are going to ask for special treatment for having done so.
Which means, for instance, that Intel may be able to field an equivalent ultralight notebook using the same supplier, but Apple made higher profit margins on the same product (although the manufacturer makes higher profits on the Intel version).
I don't have any illusions that they will continue with this level of benevolence in the future, especially after Tim (or indeed even as Tim's memory of Steve fades). I don't even know if they are still working that way now. Probably not, or not for long.
At any rate, R&D frequently is paid for by luxury goods of one form or another. Like it or not, Apple is nearly the only luxury computing maker out there, though Samsung, Sony and LG dabble. I've expressed my confusion on a number of occasions as to why things are this way, given that it's not the case with cars or a number of other goods. I like Apple products and am an investor, but I'd also like to have more opportunities to compare them to a competitor. Hyundai has figured out they need a separate luxury brand, as have a number of car makers before them. Maybe Sony, Samsung and LG need their own 'Genesis', 'Audi', "Acura".
My last example was not a case of 'poor Intel, can't sell because of a licensing agreement'. They were probably wiping their tears away with 100 dollar bills. TSMC is probably pretty happy with this arrangement too. The way they 'get out' of this exclusivity is to produce more chips than Apple cares to buy, since it's probably Right of First Refusal, not 'you can make no more than N million chips per month or else'.
Also, the shift from desktop CPUs to mobile CPUs as leading the shift to smaller nodes sounds less like a massive revelation, and more like the innevitable result of mobile CPUs out-stripping traditional CPUs in volume & margin.