Arm CEO: 50% of Windows PCs will be powered by Arm chips by 2029

midian182

Posts: 9,857   +125
Staff member
A hot potato: The announcement of the Microsoft/Qualcomm Arm-based PCs has been hailed by some as the start of a serious challenge to x86 processors' dominance. Not surprisingly, one person who believes a revolution is coming is Arm CEO Rene Haas, who thinks Arm-based chips will be powering more than 50% of Windows PCs by 2029.

Microsoft last month announced the Copilot+ AI PCs packing Qualcomm's Arm-based Snapdragon X Elite and Snapdragon X Plus SoCs. Microsoft says the Snapdragon X Elite-powered latest Surface laptop is up to 58% faster than Apple's M3 MacBook Air while offering excellent battery life, and early tests appear to confirm this – though they were paid for by Microsoft.

Windows-on-Arm PCs have been around for a while now, but the latest Snapdragon chips, which are comparable to high-end x86 CPUs, are getting plenty of attention.

In an interview with Reuters, Haas said, "Arm's market share in Windows - I think, truly, in the next five years, it could be better than 50%."

Haas added that Microsoft is very committed to Arm from a software standpoint and has gone well beyond anything it has done previously with Arm over the last couple of years.

The Arm-based Copilot+ PCs from Microsoft and other vendors are focused on consumer laptops, though Qualcomm is offering a Windows Snapdragon X dev kit in the form of a NUC-like mini desktop. Furthermore, Qualcomm CEO Cristiano Amon said at Computex that the Snapdragon X Elite and X Plus chips were coming to all form factors, suggesting future desktops powered by the SoCs. It seems Haas' prediction likely covers both laptop and desktop Windows PCs.

The new wave of Arm-based PCs is exciting, and Microsoft has gained the support of big manufacturers such as Dell, HP, Acer, Asus, Samsung, and Lenovo, but seeing x86 PCs fall below a 50% market share in just five years seems like a very optimistic forecast.

Canalys analyst Kieren Jessop told The Register that just 8 – 10% of quarterly PC shipments today are Arm-based, and almost all of those are Apple Silicon. The analyst predicts 30% of the PC market will be Arm-based by 2026, and the only way that number would be higher in a five-year time frame would be in the unlikely event of Intel introducing its own Arm chip.

Intel and AMD have just announced their AI-focused Lunar Lake and Ryzen AI 300 CPUs, respectively, as the two big x86 players try to take the spotlight off the Snapdragon chips, offering what they claim is similar or better performance than Qualcomm's offerings.

Permalink to story:

 
There is a zero percent chance that happening. Even if all x86 chips stop getting sold and you could ONLY get ARM chips for Windows from now on, with the amount of x86 device in use currently, I still don't believe you'd get a 50% market share.
 
I think he is aiming at business use. Companies want to pay as little as possible for a windows laptop and if that laptop can run all the software while on ARM then they will go for it.
 
Considering how many Windows PC's are used in corporate settings, and how slowly such environments usually get upgraded, this guy needs a reality check 😂

Windows 11 adoption, after a few years, isn't even close to the 30% mark...
 
If the chip prices for Snapdragon X are in the $150-200 range, as quoted by some people, they are well on their way to massive disruption.

Intel's low cost Meteor are in the $350 range and AMD has mostly kept older uArch for low end chips. So, if nothing changes in the next 5y, we are truly in for a lot of change in the market.
 
Shouldn't we wait for a real hands on review of these before popping the champagne corks? I don't think there's going to be a huge rush to recompile every x86 piece of software to run on arm. Microsoft has been working on WART (Windows on Arm) since the WindowsRT debacle, and I doubt they've fixed it yet, or they'd be showing it off left and right instead of hiding it by showing ARM native benchmarks.

Although it probably will carve out some market for itself. As someone mentioned already, there are some buyers who bill buy on price as long as the basic software is there, just like schools have gravitated to Chromebooks.
 
Wishful? Try delusional. To do this every Pc sold from now until 2030 would have to be ARM. And given ARM's track record on the PC space, I wouldnt hold my breath.
If the chip prices for Snapdragon X are in the $150-200 range, as quoted by some people, they are well on their way to massive disruption.

Intel's low cost Meteor are in the $350 range and AMD has mostly kept older uArch for low end chips. So, if nothing changes in the next 5y, we are truly in for a lot of change in the market.
I'll be pedantic. the one Core 3 we have so far is $295, not $350. And that is full tray price, which OEMs do not pay. We've yet to get the Pentium meteor chips, but they will come sooner or later, for lower prices.

If you want low price, the n200 is only $80. And given ARM's track record, it likely will perform about as well.
 
Considering how light our office work is using a desktop PC, an arm-based PC might be an option for our top management to buy..
well, as long as the price is right and the performance meets basic productivity needs..
 
I can't by a MB for ARM, so I can't use a discrete GPU or build my own PC. So I'm stuck with a laptop format and third rate iGPU for 1080p gaming at best.

Sure, ARM will make inroads in laptops but never in desktop. Also, ARM won't be getting a free ride in laptop either. I would take Lunar Lake and subsequently Panther Lake very seriously in performance per watt. If Intel can deliver on Lunar Lake's hype, Qualcomm is already in trouble.

The guy is so delusional it's actually hilarious.
 
There is a zero percent chance that happening. Even if all x86 chips stop getting sold and you could ONLY get ARM chips for Windows from now on, with the amount of x86 device in use currently, I still don't believe you'd get a 50% market share.
I agree. There's also compatibility issues, which are not easily solved. Microsoft can't just "flip a switch" and make everything complied and written for CISC magically run on RISC. Just doesn't work that way and no one wants it too.
 
Last edited:
Considering how many Windows PC's are used in corporate settings, and how slowly such environments usually get upgraded, this guy needs a reality check 😂

Windows 11 adoption, after a few years, isn't even close to the 30% mark...
The bigger the corporation the slower the process. And if we talk about banking sector where Win Xp is still there today. Some ATM machines still run those and are still in use.
 
They might get into laptops, especially the ones that would get whatever the equivalent of a pentium or celeron is on the low end these days, though the irony is arm's pushing has made intel and amd focus on more efficient cpu's (yes, Intel makes little hotbox cpu's for desktops and ram power into them, but stuff like the n100 is pretty efficient and quite powerful for its size and thermal output), so I don't see it, ARM is great for mobile devices, but as soon as you start pushing it to core counts and performance needed to match x86 chips, it also doesn't stack up amazingly for power efficiency, more competition is good, but a claim of 50% of pc's running arm is laughable (especially as pc's are an open ecosystem with user choice, not locked down like Apple's, so the experience will never be 100% optimised for x86 to arm translation etc.)
 
Wishful? Try delusional. To do this every Pc sold from now until 2030 would have to be ARM. And given ARM's track record on the PC space, I wouldnt hold my breath.
I'll be pedantic. the one Core 3 we have so far is $295, not $350. And that is full tray price, which OEMs do not pay. We've yet to get the Pentium meteor chips, but they will come sooner or later, for lower prices.

If you want low price, the n200 is only $80. And given ARM's track record, it likely will perform about as well.
I am going by performance standards. The cheapest MTL seems to be $332, what is priced for OEMs one doesn't know. But given the complexity, it should be an expensive chip.

Certainly you have not looked at Snapdragon Elite performance. It has nothing to do with N chips and is smashing over what Intel has right now.

Unless leaked the numbers are completely wrong, it is an M1 style game changer which even AMD will have a hard time with.
 
I think just 50% of new PCs sold by 2029 would be optimistic. I don't think Windows on ARM will have the same immediate pickup like Macs had, just because the segment is much more fragmented.
 
Unless leaked the numbers are completely wrong, it is an M1 style game changer which even AMD will have a hard time with.

M1 was not game changer at all. It was nothing else than normal mobile CPU with integrated (and non-upgradeable) memory. Leaving out Apple fanboys, nobody was interested.
 
I won't say it's impossible. I mean I expect to get an ARM for my next notebook -- but I'm running Ubuntu, and I've run "Chrubuntu" on an ARM Chromebook before; it was trouble-free, any software I wanted to install was already built for ARM. Anything in the Ubuntu repositories, and for outside stuff there was usually an ARM version as well (often "for" Raspberry Pi... but ARM is ARM, those bins would run fine on Tegra K1 in that Chromebook.) The x86/x86-64 emulation back then (this was like 4 or 5 years ago) was incomplete, now apparently you just install box86 (x86) and box64 (x86-64) and away you go if you want to run (x86/x86-64) steam or run stuff in wine or whatever

The power use! That Chromebook had a GPU about dead even with a GTX650, with OpenGL, CUDA, and Vulkan support. through the nvidia driver. Both in ChromeOS and Ubuntu I got 22 hours battery life under realistic usage; and I could peg out all 4 cores doing video encodes and still get 12 hours. That battery life was astounding.

The "Ardreno" GPU used in the Qualcomm chips is based on the mobile Radeon, and the Mesa Gallium 3D driver for it is apparently just as feature complete as the Intel and AMD GPU drivers; full OpenGL, full Vulkan including the stuff required for DX12 to work in wine and Steam/Proton.

Anyway.. I won't say it's impossible, bot the CPU and GPU in these is quite good, and battery use is very good. Perhaps Microsoft will get things sorted out well enough that people will appreciate the power savings. So I'll say it's possible. But I'd also be surprised if it went that far in terms of sales percentage.
 
It is very good idea because x86 intel and amd is very low performance very old and too much consumption
Indeed. Literally the only thing it has going for it is software compatibility. And Apple has good x86/x86-64 emulation along with the vast majority of apps being M1-native already; Linux has box86/box64 which are apparently very effective, along with a vast majority of apps being ARM-native (especially since, if you find your distro is missing something, you can often find someone built it for Raspberry Pi... that's an ARM so it'll run on your ARM Linux box.) Microsoft has good x86/x86-64 emulation and it's just a question of them getting enough ARM-native apps. I've heard this is proceeding OK.

It's like "Oh, this CPU is a bit faster than ARM!" "Yes, but look at the huge power use." "OK, this one has almost as low power draw as the ARM". "Yeah, but it's an Atom-based CPU, it's a slug." The Intel instruction set is just too complicated, you can make a simple x86-64 implementation and have it fairly small, but in that same space the ARM can have additional pipelining and etc. to make it run faster. And at the high end, well, you can make both just as fast as you want when you focus on raw speed rather than power consumption; when that speed is through massive core counts, that again goes back to wanting large numbers of simpler cores where ARM has a big advantage.
 
Back