Speculation: Intel will become fabless

Page 12 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

With the loss of its manufacturing lead, will Intel become fabless?

  • Yes, Intel is a product designer at heart, and they will seek a more flexible fabless model.

    Votes: 24 13.4%
  • No, manufacturing is integral to Intel, and they will continue to invest to stay competitive.

    Votes: 155 86.6%

  • Total voters
    179

mikegg

Golden Member
Jan 30, 2010
1,740
406
136
...and the “doom & gloom”, “x86 is dead”, and “ARM will win” continues on this forum.
There is a difference between seeing trends and proclaiming "x86 is dead". The trend is clearly pointing to more ARM in high performance computing environments. Saying this isn't saying "x86 is dead". You're being overly dramatic.

Just a while ago, no one thought ARM can even compete in the laptop space. Yet, Apple is about to blow the door open with their in-house ARM CPUs. Apple is planning to scale ARM all the way to Mac Pros. And Amazon's Graviton itself is extremely competitive in perf/cost and it's only a second-generation product.
 
  • Like
Reactions: Etain05 and amenx

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
I think this forum agrees that any Intel shift to TSMC will mean Intel has to exit the bleeding edge node competition. No one is arguing against that.

This is my personal prediction on how it's going to play out:

When do you see Intel starting this negotiation and how do you see the competitive landscape shifting between now and then?
 

mikegg

Golden Member
Jan 30, 2010
1,740
406
136
When do you see Intel starting this negotiation and how do you see the competitive landscape shifting between now and then?
Analysts believe Intel has already started.

Several sell-side analysts have speculated that such talks are already under way with Taiwan Semiconductor, which already makes chips designed by Nvidia (NVDA) and AMD.

https://www.marketwatch.com/article...lay-chip-stocks-51595626658?mod=mw_quote_news

You have to believe that Intel has already started negotiations with TSMC/Samsung or they wouldn't have talked about 3rd party fabs in their earnings call.

The competitive landscape in terms of what? Stocks? CPU performance? Market share? I've written about these in various posts in this thread already.
 

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
There is a difference between seeing trends and proclaiming "x86 is dead". The trend is clearly pointing to more ARM in high performance computing environments. Saying this isn't saying "x86 is dead". You're being overly dramatic.

Just a while ago, no one thought ARM can even compete in the laptop space. Yet, Apple is about to blow the door open with their in-house ARM CPUs. Apple is planning to scale ARM all the way to Mac Pros. And Amazon's Graviton itself is extremely competitive in perf/cost and it's only a second-generation product.

I will believe it when I see it, and even if it happens it won’t matter. Apple is a minority player with a closed platform.

Apple isn’t the only one building faster chips, and they don’t break any records when it comes to power efficiency either.

There is only one company that will sell you a 15 watt, 8 core, 16 thread SoC with a powerful GPU and it isn’t Apple.

EDIT: in case it wasn’t abundantly clear, I am saying that Apple’s engineers don’t exist in a vacuum. Both AMD and Intel are perfectly capable of competing with anything Apple puts out.
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
Analysts believe Intel has already started.

Several sell-side analysts have speculated that such talks are already under way with Taiwan Semiconductor, which already makes chips designed by Nvidia (NVDA) and AMD.

https://www.marketwatch.com/article...lay-chip-stocks-51595626658?mod=mw_quote_news

You have to believe that Intel has already started negotiations with TSMC/Samsung or they wouldn't have talked about 3rd party fabs in their earnings call.

The competitive landscape in terms of what? Stocks? CPU performance? Market share? I've written about these in various posts in this thread already.

I don't care what analysts say, the value of their words isn't worth the space it takes up in my RAM. Just 6 months ago I was readying analysts predict how Intel was going to regain tech leadership with their new 10 nm CPUs and would leave AMD in the dust. That didn't age very well.

I'm asking you when you think that will happen since you seem to think whenever it does happen, it will be a huge negative for AMD. Edit: I don't think them mentioning 3rd party fabs is any indication that negotiations have begun with other fabs. They easily could have just been saying that to give the impression that they have a plan no matter what happens, doesn't mean they actually have an actionable plan.

Competitive landscape in terms of market share and CPU performance.
 

mikegg

Golden Member
Jan 30, 2010
1,740
406
136
I will believe it when I see it, and even if it happens it won’t matter. Apple is a minority player with a closed platform.
This isn't the year 2005 anymore. It's 2020 and Apple is the most valuable company in the U.S. Apple has huge pull in the industry.

I don't think you're quite understanding the ramification of Apple's move to ARM. Like you said, Apple doesn't exist in a vacuum.

1. Apple moving to ARM means more professional software will support ARM, which allows Windows ARM to grow.
2. Apple successfully moving to ARM means PC makers will have to consider ARM options in order to compete.


Apple isn’t the only one building faster chips, and they don’t break any records when it comes to power efficiency either.
But they have broken records when it comes to power efficiency.

There is only one company that will sell you a 15 watt, 8 core, 16 thread SoC with a powerful GPU and it isn’t Apple.
For now.

Just 6 months ago I was readying analysts predict how Intel was going to regain tech leadership with their new 10 nm CPUs and would leave AMD in the dust.
No one believed that. I sure as hell didn't believe that. However, most people here are now finally agreeing that Intel must become fabless in order to compete again.

I'm asking you when you think that will happen since you seem to think whenever it does happen, it will be a huge negative for AMD.
When will it happen? My guess is 2-3 years. That's just a guess.

Edit: I don't think them mentioning 3rd party fabs is any indication that negotiations have begun with other fabs. They easily could have just been saying that to give the impression that they have a plan no matter what happens, doesn't mean they actually have an actionable plan.
Or maybe they're not as stupid as we think. After all, those Intel executives are quite a bit more accomplished than the average Anandtech poster.

My bet is that they're seeing the same thing as everyone is seeing: they've lost the node advantage forever. If they ever want to compete for the performance crown again, they must go fabless.

One of the reasons why they hired Jim Keller was so that he can lead the effort to decouple their process and design teams. Intel wanted to do that, years ago.

Competitive landscape in terms of market share and CPU performance.
In the next 2-3 years, I expect Intel to continue to lose market share in servers, laptops, and DIY desktop. I expect AMD to continue to outpace Intel in multicore performance and eventually equal Intel in single-core performance.
 
Last edited:
  • Like
Reactions: Spartak

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
This isn't the year 2005 anymore. It's 2020 and Apple is the most valuable company in the U.S. Apple has huge pull in the industry.

I don't think you're quite understanding the ramification of Apple's move to ARM. Like you said, Apple doesn't exist in a vacuum.

1. Apple moving to ARM means more professional software will support ARM, which allows Windows ARM to grow.
2. Apple successfully moving to ARM means PC makers will have to consider ARM options in order to compete.



But they have broken records when it comes to power efficiency.


For now.


No one believed that. I sure as hell didn't believe that. However, most people here are now finally agreeing that Intel must become fabless in order to compete again.


When will it happen? My guess is 2-3 years. That's just a guess.


Or maybe they're not as stupid as we think. After all, those Intel executives are quite a bit more accomplished than the average Anandtech poster.

My bet is that they're seeing the same thing as everyone is seeing: they've lots the node advantage forever. If they ever want to compete for the performance crown again, they must go fabless.

One of the reasons why they hired Jim Keller was so that he can lead the effort to decouple their process and design teams. Intel wanted to do that, years ago.


In the next 2-3 years, I expect Intel to continue to lose market share in servers, laptops, and DIY desktop. I expect AMD to continue to outpace Intel in multicore performance and eventually equal Intel in single-core performance.

Apple has negligible marketshare on the desktop, and pretty much zero percent in the racks. They are irrelevant except that they caused Intel pain and shame. They don’t matter to this discussion except to the point that they put Intel in a tough spot.

Right now everyone is hyping Apple’s future moves, but they have yet to even show they can execute. The Mac line has existed for many decades and STILL doesn’t have significant marketshare.

Apple didn’t leave Intel because Intel couldn’t deliver. They left Intel because they wanted total control and total profitability.
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
No one believed that. I sure as hell didn't believe that. However, most people here are now finally agreeing that Intel must become fabless in order to compete again.

I was just pointing out that "analysts" aren't really a good place to turn to for understanding really anything in this industry. They are wrong far more often than they are right because almost none of them have any kind of technical background to actually understand what is happening.

When will it happen? My guess is 2-3 years. That's just a guess.

If it is to happen in 2 - 3 years then Intel indeed needs to be negotiating now with TSMC/Samsung, stopping R&D on future nodes, looking at how to shutter their own fabs or turn them into 3rd party fabs, and TSMC/Samsung need to start building additional fabs very, very soon to meet demand. All that I've seen and heard doesn't make any of this seem realistic.

Or maybe they're not as stupid as we think. After all, those Intel executives are quite a bit more accomplished than the average Anandtech poster.

You could have easily said the same thing about 10 nm and yet. . .

My bet is that they're seeing the same thing as everyone is seeing: they've lots the node advantage forever. If they ever want to compete for the performance crown again, they must go fabless.

One of the reasons why they hired Jim Keller was so that he can lead the effort to decouple their process and design teams. Intel wanted to do that, years ago.

I don't think many at Intel believe their fab is not forever behind in the race and honestly, it's not a sure thing at this point but it's looking more and more that way without something to shakeup what's happening there.

As far as decoupling design and process, Intel wanted to do that years ago? According to whom? If they wanted to do it years ago, why didn't they do it? Did they really have to wait exclusively for Keller to become available before making this change?


In the next 2-3 years, I expect Intel to continue to lose market share in servers, laptops, and DIY desktop. I expect AMD to continue to outpace Intel in multicore performance and eventually equal Intel in single-core performance.

Ok, so let's assume for a moment Intel is planning this move in 3 years. If Intel continues to lose market share in every CPU segment, how long before AMD with CPU, GPU, and semi-custom designs becomes a more valuable customer to TSMC compared to a potential Intel customer? If this happens, who is more likely to be on the leading node and who is more likely to be stuck 1 - 1.5 nodes behind?
 
  • Like
Reactions: Tlh97 and scannall

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
1. Apple moving to ARM means more professional software will support ARM, which allows Windows ARM to grow.
2. Apple successfully moving to ARM means PC makers will have to consider ARM options in order to compete.

I agree with your first point, but not your second. What does Apple being on ARM have to do with PC OEMs wanting to stick with tried and true x86 versus going out on a limb with much less well supported (in terms of Windows software) ARM? They can't buy Apple's SoCs, so if they perform really well it doesn't give PC OEMs new options.

Besides, just like with phones people don't choose between macOS and Windows based on performance. If you have a Windows PC today, and Apple ships a Mac that's 20% faster than any PC you can buy, you aren't going to buy a Mac because of that unless you are a statistician or some other narrow field where absolute performance matters more than software compatibility. Even a Mac that was twice as fast as a PC wouldn't get too many people to switch - gamers need a platform that has the games they want to play. Businesses need something that can run the Windows software they require, and so on.

All Apple successfully moving to ARM will do to the PC market is help make Windows/ARM a bit more viable because all the ARM Mac owners who want to run Windows will be using it instead of Windows/x86. It isn't going to change the calculations for OEMs one bit. Heck, Apple leaving x86 means they probably think about Apple LESS than they do now - at least currently PC OEMs are nominally competing with Apple because someone who thinks Apple makes the best x86 laptop might buy them to run Windows 100% of the time rather than buying from Dell etc. Once they've gone ARM that will no longer be possible.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
Apple didn’t leave Intel because Intel couldn’t deliver. They left Intel because they wanted total control and total profitability.

You're wrong. Apple has called Intel out (not by name, but by inference when mentioning stuff like "chip suppliers") several times in their calls over the past few years for their failures to deliver. One of the reasons a company wants "total control" is if their suppliers are unable to execute. Everyone knows Apple is very demanding of its suppliers, and not very forgiving of mistakes that impact Apple's schedules.

How many times has Intel f'ed up in the last five years? You think Apple didn't notice or care? Who knows what sort of fairy tales they've been feeding Apple and their other big customers about how well 10nm was doing, and how it was right around the corner? Because from everything we've seen, it seems like the problem was not that Intel was deliberately lying to others when they painted a rosy picture of 10nm - they were lying to themselves because they are so screwed up and so many people are unable or unwilling to be the bearer of bad news that they don't actually know how bad things are.

As for profitability, we'll have to see what type of solutions Apple releases. I think Apple may well have given their engineers a lot of leeway to develop the best possible ARM SoCs they can for the Mac to beat Intel's performance in every metric, to silence all the critics who are claiming Macs will become inferior due to the switch to ARM. If that costs more than what they were paying Intel (at least at first) then so be it.
 

mikegg

Golden Member
Jan 30, 2010
1,740
406
136
I agree with your first point, but not your second. What does Apple being on ARM have to do with PC OEMs wanting to stick with tried and true x86 versus going out on a limb with much less well supported (in terms of Windows software) ARM? They can't buy Apple's SoCs, so if they perform really well it doesn't give PC OEMs new options.
Apple demonstrating that you can have more performance than an x86 computer (particularly in the laptop space) while having more than double the battery life, is a game-changer that other PC makers can't ignore.

I expect Apple to take the performance and battery life crown in the laptop segment.

Next, I expect Apple to start creeping into the large $700-$1000 laptop range and be able to offer more performance and more battery life than anything PC makers can offer here, which they have never been able to do before with Intel and POWER processors

A lot of buyers want a Mac but Intel Macs have been out of reach for them due to the price. With ARM Apple can finally start to penetrate this market, using the same strategy as the iPhone SE.

People are and should be comfortable with owning a Mac. iPhone owns ~40-50% of the U.S. market share. iPad 65%. They're already familiar with Apple's designs and walled garden. Up until now, Apple couldn't make a sub $1,000 laptop that wasn't crap (said by Steve Jobs himself). They will be able to with ARM.

I fully expect Macs to accelerate in market share once ARM Macs start releasing.

When that happens, combined with more professional software makers optimizing for ARM, I expect PC makers to start shipping more and more Windows/Google ARM computers.

I expect Qualcomm to benefit. I expect a repeat of iOS vs Android in which Qualcomm dominates Android/Windows ARM segment.

I expect Qualcomm and other ARM designers to put much more effort into laptop and desktop ARM chips after Apple's transition, thus accelerating the competition between ARM and x86 in laptops and desktops.

It's true that PC gamers and enthusiasts, which happen to be the majority on Anandtech forums, will always choose a Windows machine, no matter how powerful an Apple ARM SoC is. But this market is small and not representative of the overall market.

ARM Macs will have a drastic effect on PCs.
 
Last edited:
  • Like
Reactions: teejee and scannall

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
Apple demonstrating that you can have more performance than an x86 computer (particularly in the laptop space) while having more than double the battery life, is a game-changer that other PC makers can't ignore.

I expect Apple to take the performance and battery life crown in the laptop segment.

I expect Apple to do so as well, but if Apple's ARM Macs have better performance and better battery life it won't be because they are 'ARM', it will be because Apple's designers (combined with TSMC's process) produced a superior product.

People need to get over this idea that ARM has some sort of inherent advantage in power, or that x86 has some sort of inherent advantage in performance. Neither do. The days when CISC vs RISC mattered are a quarter century behind us. Transistor budgets are so large the additional decoding to handle x86 is a rounding error for power draw.

The reason why ARM designs have done better than x86 power wise is because that's what they've been designed for. ARM started out in embedded, where power use is often the most critical factor. Then they moved into mobile devices, where power use is still of paramount importance.

x86 started out in PCs, 100% of which were plugged into the wall and didn't care about power draw at all. Power didn't start mattering until two decades later when laptops became a significant segment of overall PC sales, but even then the designs had to be a compromise between power and performance.

That's why ARM designs have traditionally done better power-wise, and why x86 designs have traditionally done better performance wise. Has nothing whatsoever to do with any differences in the two ISAs.
 

mikegg

Golden Member
Jan 30, 2010
1,740
406
136
I expect Apple to do so as well, but if Apple's ARM Macs have better performance and better battery life it won't be because they are 'ARM', it will be because Apple's designers (combined with TSMC's process) produced a superior product.
Samsung used Qualcomm's Snapdragon for a Windows ARM laptop and achieved 25 hours of battery life in an ultra-thin laptop.


It's too bad that Windows ARM is quite bad and doesn't have support from software makers. In addition, Samsung is using a mobile chip (Edit: a low effort desktop version of a mobile chip) in a laptop because Qualcomm doesn't make a dedicated laptop chip.

I'm betting that both of these issues will be solved when Apple makes the transition:

1. Software makers will have to support ARM, allowing Windows ARM to become viable
2. CPU makers, in particular Qualcomm, will take the laptop/desktop market more seriously and start making dedicated PC ARM chips.

This is a chicken and egg problem. Apple will be the egg.

Anyway, I think we've taken this thread off the rails. This will be my last post on ARM vs x86, Macs, etc.
 
Last edited:
  • Like
Reactions: Gideon

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
Apple has never delivered a chip that has outperformed x86. I wouldn’t get your hopes up that they will do it now. What they will deliver is reasonable performance with decent battery life. However, this thread isn’t about Apple, it is about Intel, or more specifically, their fabs.

If I were Intel, I would consider opening things up to others. It would require additional investment, but it would help them increase capacity and decrease the dependence on their CPUs for volume and profitability.
 
  • Haha
Reactions: Spartak

Failnaught

Member
Aug 4, 2008
26
25
91
I recently asked an Intel recruiter what they are going to do to recapture leadership status in fab process. The engineer said it's really a matter of which metric you pick, and so they're not really behind at all (basically Intel 10nm = TSMC 7). It was the wrong forum for this kind of question, to be honest, and the recruiter wasn't involved in fab when he was previously working in more technical roles, but this kind of answer makes me very nervous. Either he is putting on a brave face at a recruitment event (very possible) or TSMC's current superiority hasn't sunk in yet in the Intel culture.
 

chrisjames61

Senior member
Dec 31, 2013
721
446
136
Apple has negligible marketshare on the desktop, and pretty much zero percent in the racks. They are irrelevant except that they caused Intel pain and shame. They don’t matter to this discussion except to the point that they put Intel in a tough spot.

Right now everyone is hyping Apple’s future moves, but they have yet to even show they can execute. The Mac line has existed for many decades and STILL doesn’t have significant marketshare.

Apple didn’t leave Intel because Intel couldn’t deliver. They left Intel because they wanted total control and total profitability.
Sorry, but the guy is right. You just don't get the ramifications of Apple moving their product line. Others will slowly start following suit and eventually their will be a critical mass towards ARM and away from X86. It may take 5 to 10 years. The economies of scale will be in ARM's favor just like how X86 squeezed out 68K and PPC.
 

chrisjames61

Senior member
Dec 31, 2013
721
446
136
Apple didn’t leave Intel because Intel couldn’t deliver.

Are you kidding? It has been reported for years on web sites just as this that Apple had to delay refreshes to Macs for extended periods of time due to problems Intel had delivering on their road map.
 

LightningZ71

Golden Member
Mar 10, 2017
1,627
1,898
136
As it relates to x86 volume on Intel's fabs diminishing, I see a larger shift in the whole software market to platform agnostic web based applications. Most of those things are just dressed up front ends to "cloud" applications at some virtual host far away. Most of that stuff can run on ANY processor that can operate a standards compliant web browser with the relevant plug ins. This one thing accounts for the use case for so many business PCs. In the blink of an eye, many, many of them could be swapped out to arm based systems with the end user likely never knowing anything had changed under the hood.

90% of the PC market could be satisfied with the SoC that's running inside of the average $250 smart phone. The only ones that REALLY care are the professionals that use a very specific set of programs and gamers that want the absolute best gaming performance that they can get. A lot of those specific software packages can be redeveloped for ARM and, with a decent enough SoC, perform just as well as they do now in many cases.

Apple hasn't penetrated the market more than they have because they have intentionally priced themselves out of 90% of the market. Period. Most users of cheap PCs would LOVE to have a Mac as most see them as more of a status symbol than an actual tool. That was by design.

Unless Intel can get into using their fab space for arm products in the future, on competitive nodes, I can't see a rosy picture for them in the following decade.
 

mikegg

Golden Member
Jan 30, 2010
1,740
406
136
As it relates to x86 volume on Intel's fabs diminishing, I see a larger shift in the whole software market to platform agnostic web based applications.
Yep. Source: I'm a web app developer. Chrome OS has demonstrated that most people can use mostly web apps fine.

Apple hasn't penetrated the market more than they have because they have intentionally priced themselves out of 90% of the market.
I do believe Apple truly thought they couldn't make a good sub $1,000 computer. I think this will change because a sub $1,000 Mac with ARM will still have amazing battery life and be competitive with x86 laptops in performance.

Most users of cheap PCs would LOVE to have a Mac as most see them as more of a status symbol than an actual tool.
Also true. Though I also believe that people genuinely want to pair their iPhones with Macs and stay in the ecosystem.

My take:

A lot of buyers want a Mac but Intel Macs have been out of reach for them due to the price. With ARM Apple can finally start to penetrate this market, using the same strategy as the iPhone SE.

People are and should be comfortable with owning a Mac. iPhone owns ~40-50% of the U.S. market share. iPad 65%. They're already familiar with Apple's designs and walled garden. Up until now, Apple couldn't make a sub $1,000 laptop that wasn't crap (said by Steve Jobs himself). They will be able to with ARM.


Unless Intel can get into using their fab space for arm products in the future, on competitive nodes, I can't see a rosy picture for them in the following decade.
It's not going to be as rosy as the last decade. In the last decade, Intel had zero competition in the cloud, laptop, and desktop space. Zero. I'm pretty sure could have driven AMD out of business but they didn't want to attract antitrust attention.
 
Last edited:
  • Like
Reactions: Tlh97

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
How reliable is this website?

The key takeaways that hexus list are this:

  • Intel has reached an agreement with TSMC
  • TSMC will begin mass production of Intel CPUs and/or GPUs next year
  • Intel chips will be fabricated on TSMC's 7nm optimised version of its 6nm process. (I'm not sure if that means TSMC N7P, N7+, or N6.)
Which fits perfectly with what we already know, that Intel will be using TSMC 7nm (maybe 6 I guess) for Ponte Vecchio next year. There is no chance Intel moves over their CPUs to TSMC by next year.