PC CPUs are getting more interesting, and competition is coming

Jay Goldberg

Posts: 68   +1
Staff
Why it matters: Five years ago there were only two companies that made CPUs, today there are a dozen. Most of the new entrants went after the big, profitable data center market, but now competitors are coming for PCs. Nvidia and AMD are reportedly preparing Arm-based CPUs for PCs. With Microsoft opening up the market for Arm laptop CPUs, this spells bad news for Qualcomm today, and potentially bad news for Intel over the very long term.

As much as Intel has struggled in the data center for the past five years, they have managed to hold onto share in PCs. These products do not offer as rich margins as data center CPUs, but they are significant volume and go a long way to keeping Intel's fabs utilized, and thus viable.

Editor's Note:
Guest author Jonathan Goldberg is the founder of D2D Advisory, a multi-functional consulting firm. Jonathan has developed growth strategies and alliances for companies in the mobile, networking, gaming, and software industries.

Intel has held onto its PC share largely due to two factors: the Intel brand and "channel control." Consumers do not care about, or are even much aware of, semiconductor production processes or instruction set architectures, but they do know the Intel brand, the result of the company's multi-decade advertising spending.

Selecting a PC CPU, for most consumers, is a maze of impenetrable specs, and so even if the latest AMD CPU was better on paper than the competing Intel CPU, Intel could still win out. Secondly, consumers are not buying from Intel, they are buying from one of the PC brands – HP, Dell, Asus, Lenovo, etc. Those companies are tightly coupled to Intel, in no small part due to the marketing payments they receive from Intel which make up the bulk of their PC profitability. Those companies are loathe to move too far away from Intel for fear of losing out on those subsidies.

In fact, the only company to enter the PC CPU market in recent years is Qualcomm. Qualcomm has been working for almost a decade to prise open a share of the market with its Arm-based CPUs. This took a lot of work, porting Windows to Arm was never going to be easy, and the relationship between Qualcomm and Microsoft has been strained as a result. That said, Qualcomm now seems to have a fairly competitive CPU.

We have written a bit about Qualcomm's efforts here, and the summary is that it is unlikely Qualcomm can win meaningful share in this market any time soon, unless on-device AI support soon becomes very important to consumers. This seems unlikely, but Qualcomm seems to have the best laptop-grade AI cores (aka, NPU) on the market for the moment.

Qualcomm has come under intense heat for its long investment in PCs, not least from us. However, one of the key selling points Qualcomm had working in its favor was the exclusive nature of its relationship with Microsoft. For all these years, Qualcomm was the only company Microsoft worked with to port Windows to Arm. Which is why the Reuters story about Nvidia and AMD Arm CPUs is so important.

Microsoft has moved on and will support other vendors' CPUs for Windows, ending Qualcomm's exclusivity. Critically, both of the new entrants have much deeper Windows roots than Qualcomm. Qualcomm has clearly struggled to get the software ecosystem around Arm-Windows up and running, and we have to suspect that the new entrants will have a much easier time of it, in some part due to the work that Qualcomm has already done. We have to think that Qualcomm will now revisit its efforts in PCs. A small share of a market that is not growing, but which has suddenly become more competitive – this is not the kind of markets where Qualcomm excels.

And what are we to make of Nvidia and AMD's efforts? AMD's entrance frankly just confuses us. They already have a decent market share in PC laptops, but decent in the sense that they have always been a distant #2 to Intel despite years of work and lots of Intel stumbles. We have to imagine they did this in part as a goodwill effort to their long-time partner Microsoft (something that Qualcomm needs to learn how to do).

Is AMD really going to spend serious marketing dollars to win share here, when there is a good chance they will just be cannibalizing their own share? What is the average consumer going to do when they compare an AMD x86 laptop to an AMD Arm laptop? They are going to be confused is what, and probably move down the shelf to the Intel brand they know.

On the other hand, Nvidia has at least a credible case. They already have a strong consumer brand, admittedly one centered on gaming, but that counts for a lot. They also get the benefit of doing a favor for Microsoft, a big customer and partner, and they have no competing product to cannibalize. We could even make the case that they will invest in marketing here because an Nvidia CPU/GPU combination laptop could be a real product category. We know many gamers who would probably flock to pick one up.

All of which brings us to the real topic of this piece – Apple. We are constantly surprised that in so much discussion of PC CPUs the semis vendors avoid mentioning Apple – a giant blind spot. We have even heard Intel executives claim that "we do not compete with Apple."

Apple has steadily chipped away at PC market share, and most critically at PC profitability share. The average Windows PC sells for at least $500 less than the lowest priced Mac. Apple consumes the vast bulk of personal computing profitability, just as they do in mobile phones. We have not crunched the data lately, but we are fairly certain that shift to Apple silicon's M1 CPU has only exacerbated this gap. This problem is so large for the other laptop makers that it is almost easier not to think about it.

Microsoft is keenly aware of the issue, and while their fortunes do not depend on the PC market so much, it is still a large, important market for them both in terms of profitability and broader strategy. They need an answer to the lack of PC profitability and seem to have landed on the CPU as an important element in their strategy. There is some logic to this, the M-powered MacBooks enjoy a reputation for being more power efficient than Windows laptops. However, we know many in the Windows supply chain who seem to fetishize Arm CPUs – if only we had Arm CPUs we could compete better with Apple. We think this misses the mark badly. Apple succeeds because it can tie its software to its silicon so tightly – the Arm part is not the differentiation.

Be that as it may, the new Windows CPUs may inject some life into the market. Arm-based chips succeeded in mobile in large part because of the many semis vendors competing here. Arm makes a big deal about its mobile ecosystem (in its IPO prospectus, for instance), all those companies competing led to faster innovation and advancement.

Also read: Intel boss Pat Gelsinger calls Arm's PC threat "insignificant"

This may come to pass in laptops, too, especially if Nvidia and AMD are just the first entrants. A healthy Arm-Windows CPU ecosystem has the potential to spark a new round of innovation in Windows laptops, by allowing deeper segmentation. Nvidia can take the high-end with expensive gaming laptops, AMD and Qualcomm can find their niches, someone will take the low-priced approach and give Google Chromebooks a run for their money.

Too soon to tell if this will actually come to pass, but the possibility now exists. For the moment, we think Intel is fairly sheltered, especially as it starts to bring better products to market with its improved manufacturing. Over the longer term though, if we start to see healthy competition from the Arm CPU makers, Intel could face a real challenge.

Permalink to story.

 
The article wrote: "AMD's entrance frankly just confuses us. They already have a decent market share in PC laptops, but decent in the sense that they have always been a distant #2 to Intel despite years of work and lots of Intel stumbles. "

AMD's growing laptop share is precisely BECAUSE they have worked hard. Not DESPITE years of work, or because Intel stumbled.

The truth is that AMD succeeded despite Intel's extremely dishonest business practices, for which they were fined over a billion dollars. I will never, ever purchase another Intel product. As a moral human being, there are certain lines I will not cross. (LOL)
 
The article wrote: "AMD's entrance frankly just confuses us. They already have a decent market share in PC laptops, but decent in the sense that they have always been a distant #2 to Intel despite years of work and lots of Intel stumbles. "

AMD's growing laptop share is precisely BECAUSE they have worked hard. Not DESPITE years of work, or because Intel stumbled.

The truth is that AMD succeeded despite Intel's extremely dishonest business practices, for which they were fined over a billion dollars. I will never, ever purchase another Intel product. As a moral human being, there are certain lines I will not cross. (LOL)
They are referring to arm pcs not x86. That's why they said the entrance confuses us. Nothing to do with laptops in general.
Nothing in business is what it seems. No company plays fair, business is war, make as much money as you can. Intel and Samsung used PR to get to the top. AMD had blunder after blunder, Intel just happened to capitalize on it. No company has clean hands. They have all done shady practices, some just get more scrutinized than others.
 
No thanks, ARM brings nothing to the table.

Where there is really a lack of competition is in chip factories, TSMC is swimming ahead alone for about 3-4 years. I'm rooting for Samsung, the precedents make me suspicious of Intel's words.
 
It was hilarious to see that clown Gelsinger just dismiss ARM out of hand last week. He said they are of no consequence. Pride cometh before the fall. Keep it up Pat, we need some light entertainment.
 
AMD already has a license to make x86 CPUs.
They're vastly more popular among Windows users, since they run most existing software natively.
So why would AMD branch out into making ARM CPUs for Windows PCs? To gain some minimal amount of market share?
I should think the answer is obvious. To protect their market share in case, for some reason, the market switches to ARM CPUs for Windows. Better to cannibalize your own x86 market share than to have it eaten by other ARM vendors. Sure, at present this seems like an extremely unlikely event, but the future is unpredictable, and it's better to have possibilities covered than to be caught unprepared.
 
ARM on PC will go nowhere so long as MS has such terrible translator performance and linux has effectively nothing. I hope this changes, I'd love to have strond M3 competition int he PC space.
No thanks, ARM brings nothing to the table.

Where there is really a lack of competition is in chip factories, TSMC is swimming ahead alone for about 3-4 years. I'm rooting for Samsung, the precedents make me suspicious of Intel's words.
Well they bring far superior battery life.
 
The article wrote: "AMD's entrance frankly just confuses us. They already have a decent market share in PC laptops, but decent in the sense that they have always been a distant #2 to Intel despite years of work and lots of Intel stumbles. "

AMD's growing laptop share is precisely BECAUSE they have worked hard. Not DESPITE years of work, or because Intel stumbled.

The truth is that AMD succeeded despite Intel's extremely dishonest business practices, for which they were fined over a billion dollars. I will never, ever purchase another Intel product. As a moral human being, there are certain lines I will not cross. (LOL)
They are all dishonest. No atheists in foxholes and no integrity in business.
 
I mean.... Honestly the more competition there is on the market, the best for the average consumer. I would like Intel to finally wake up and go back to the core 2 duo, core 2 quad era, but the CPU that I have right now (i9 9900k) is probably going to be the last one I buy from them unless I see something good in about 2 to 3 years from them.
 
"What is the average consumer going to do when they compare an AMD x86 laptop to an AMD Arm laptop? They are going to be confused is what, and probably move down the shelf to the Intel brand they know."

A lot of speculation here

The a lot of more young savvy people will be buying - maybe by then can have a dual OS - android and windows
If by average you mean ignorant- they will just asked the sales - person

The ARM will probably be sold as a Chromebook competitor - long battery life - my useless speculation

Why would Nvidia be a credible case - what credible case as a CPU manufacturer - do you think average Joe knows about Tegra chips ?

Apple had both

Don't confuse certain consumers from certain countries with those from say Sweden , Germany , Singapore

AMDs position makes perfect sense - they are gaining knowledge in a wide variety of chips
I have stated here again and again the likes of AMD would like to create fit for purpose huge SOCS - chiplets , x86, ARM , FPGA- . AI addons , GPU

I don't think ARM will beat x86 - x86 beat ARM

Machinery , sensors , robots , Vehicles , etc are coming to have more and more chips
why get stuck without a big player like ARM in your arsenal

GPUs are still a bit limited in being able to create a fully fleshed out worlds - with independent agents , environments

You have APUs - great as don't need a GPU to do system checks, get a display- run a functional windows

Why not GPUs and a ARM CPU attached - could run in with or without a dedicated CPU
Gamewise that ARM CPU on the GPU could do a lot of processing very quickly , that does need to go to CPU/memory and back

Maybe that's why Nvidia and AMD are interested as they feel their dedicated cards whether AI or GPU need this to evolve

probably much easier to have an array of 256 ARM cpus than the same number of x86 cpus

Power is probably lower , are cooler to run

 
ARM on PC will go nowhere so long as MS has such terrible translator performance and linux has effectively nothing. I hope this changes, I'd love to have strond M3 competition int he PC space.

Well they bring far superior battery life.
I don't believe that, considering models with similar specs including battery capacity, I see AMD laptops that have autonomy and efficiency similar to M2 chips (outside the Asics camp.), I believe that both AMD and Intel will start using ASICs to speed up specific tasks, which is what apple did.
 
Well they bring far superior battery life.
ARM is only more efficient because it was designed that way. That is, x86 could be as efficient if designed with the same priorities.

But the current advantage is good as the competition will spur AMD and (to a lesser extent) Intel to focus on efficiency in addition to the performance crown.
 
1 ISA is irrelevant to the battle for performance per watt. There is work that needs to be done by Intel and AMD, but achieving the performance/power curve of Apple Silicon and beyond can be done with any and all ISAs. Instruction decode is a miniscule component of modern CPU power draw.

2. The problem for Qualcomm is they are missing the most important aspect of Apple’s success. Apple’s efficiency cores are the real star of Apple’s day to day usage battery success. The performance core work per watt is a tremendous achievement, but the efficiency core work to watt is extraordinary. 1/3 the P-core work for 1/8 the power. It allows Mac laptops to spend a lot of time with all performance cores off using only somewhat more power that an Apple Watch while filling the user’s needs.

Qualcomm will need an effective low power core regardless whether their Oryon core is the best or not. Notice despite all the performance/watt talk, they made no claims about device battery life.

3. By the time Qualcomm ships all the other major players will have on-board AI accelerators. They won’t be able to stand out there.

So, what is the selling point? I don’t see any breakthrough feature.
 
AMD holds sub 20% market share in every market but Client. EVGA didn't go further than selling GPUs while their competition spread out to other markets.

Point being, if you stand still, you'll get left behind.

Qualcomm Unveils Even More Snapdragon X Elite PC CPU Benchmarks: 23W & 80W Reference Laptops Tested
 
Last edited:
The truth is that AMD succeeded despite Intel's extremely dishonest business practices, for which they were fined over a billion dollars. I will never, ever purchase another Intel product. As a moral human being, there are certain lines I will not cross. (LOL)
Like when AMD tried to lock Zen 3 to 500 series boards? Would you have stuck with them anyway or would you have forgiven Intel?

When does that $1.25B Excuse Card expire btw?

AMD agreed to drop a lawsuit against Intel in Delaware and two cases in Japan, saying the settlement creates a level playing field and is “a pivot from war to peace.” AMD noted though that some narrow issues on Intel rebates remain.

The two companies also sealed a five-year cross license deal and gave up any claims of breach from a previous licensing agreement, paving the way to make AMD fully “fabless.”
 
Last edited:
AMD holds sub 20% market share in every market but Client. EVGA didn't go further than selling GPUs while their competition spread out to other markets.

Point being, if you stand still, you'll get left behind.

Qualcomm Unveils Even More Snapdragon X Elite PC CPU Benchmarks: 23W & 80W Reference Laptops Tested
The answer is that it can't compete with AMD, x86 has something special called out-of-the-box compatibility built in decades not days. Apple has a special status, they control the entire ecosystem from software to hardware and have a captive audience on a leash, so no matter what's running under the hood or how limited you are, these people are going to buy what Apple sells.

I just wonder how massive this chip is that Qualcomm created in hopes of competing w, my bet is that it must be the size of a server chip and just as expensive to produce.
 
I don't believe that, considering models with similar specs including battery capacity, I see AMD laptops that have autonomy and efficiency similar to M2 chips (outside the Asics camp.), I believe that both AMD and Intel will start using ASICs to speed up specific tasks, which is what apple did.
Show me a windows laptop with a 100w combined TDP that manages 19 hours of streaming on a charge. I'll wait.
ARM is only more efficient because it was designed that way. That is, x86 could be as efficient if designed with the same priorities.

But the current advantage is good as the competition will spur AMD and (to a lesser extent) Intel to focus on efficiency in addition to the performance crown.
It could. And a unicorn could appear in my shower tonight too.

We've never seen an x86 chip that can match ARM's efficiency and performance at the same time, and if nobody makes one then it is a totally moot point.
 
Show me a windows laptop with a 100w combined TDP that manages 19 hours of streaming on a charge. I'll wait.
It could. And a unicorn could appear in my shower tonight too.

We've never seen an x86 chip that can match ARM's efficiency and performance at the same time, and if nobody makes one then it is a totally moot point.
Do you mean streaming = watching video online Or streaming = video/live transmission ?

LG gram 14Z90P, HP dragonfly G4 and some ultraportable Lenovo Yoga reach 15-20hrs browsing or watching videos. The only way to be able to transmit video with quality for such a long time must be through ASICs, this is not a merit of the ARM architecture and it is not the battery life that you get under high real load, In this case, the M2 pro on the very expensive Macbook completely drains the battery in 70 minutes, being inferior to the x86 models mentioned.
 
The answer is that it can't compete with AMD, x86 has something special called out-of-the-box compatibility built in decades not days. Apple has a special status, they control the entire ecosystem from software to hardware and have a captive audience on a leash, so no matter what's running under the hood or how limited you are, these people are going to buy what Apple sells.

I just wonder how massive this chip is that Qualcomm created in hopes of competing w, my bet is that it must be the size of a server chip and just as expensive to produce.
Not sure what the rush is. Everyone starts at the bottom.
 
It could. And a unicorn could appear in my shower tonight too.

We've never seen an x86 chip that can match ARM's efficiency and performance at the same time, and if nobody makes one then it is a totally moot point.
It is not a moot point. It's an important point that ARM doesn't have some magic efficiency advantage over x86. So a revamped x86 with more focus could achieve similar efficiency. Similarly as ARM (including Apple) push up their performance focus the efficiency gap is diminishing.

Again the point is that the design itself not ARM vs x86 is where the efficiency is coming from.
 
Back