For the purpose of this discussion we will peg this as the static number of PCs shipped annually for the next 10 years.
For reference PC shipments improved to >275 million worldwide shipments in 2020.
This was before Apple left AMD/Intel with 22.5 million worldwide shipments of Macs for Apple Silicon in Nov 2020.
Leaving 252.5 million Intel/AMD PCs left.
I forecast Expect >80% of all x86 PCs becoming SoC by 2030s.
This would make >202 million PCs with SoC & less than <50.5 million PCs with non-SoCs.
Why? Because Apple's doing it and the PC industry will see it as the easier/cheaper/better/profitable way to make laptops/desktops for the mass market while improving performance per watt, power consumption, battery life & raw performance. This would reduce the complexities of PC's supply chain.
This will negatively impact users who prefer frictionless parts replacements as global units shipped will drop.
I can imagine new Nvida, AMD & Intel CPU & dGPU parts to be released further apart by a few more quarters to even years when volume drops to <20% or <50.5 million.
Nvidia, AMD & Intel SoC parts will increase in volume and may receive far more frequency in newer/faster/better SKUs.
Although AMD/Intel's market share may shrink even further when Windows 11 on ARM will finally address fat binaries & dynamic binary translation as well as those on the Mac.
I could imagine that Android SoC makers Mediatek, Qualcomm, HiSilicon (Huawei) and Samsung will want to expand into the Windows 11 on ARM PC space.
In terms of performance per watt they're between Apple silicon and x86.
Over 1 billion Android smartphones are shipped annually.
So them pursuing the quarter billion PC market would be easy for them to execute. Like Apple most of the R&D cost for PC SoC part has already been paid for by smartphone sales so a little bit more for a SKU specific to PCs wouldn't cost as much as for AMD/Intel.
Conservatively ARM SoC PCs may eat into 20% of 252.5 million PCs.
Leaving Intel/AMD with 202 million PCs for both SoC & non-SoC.
With such a smaller PC market >161.6 million of all x86 PCs will be SoC & <40.4 million of all x86 PCs will be non-SoC.
<40.4 million of the frictionless parts replacement PCs we enjoy today will really be something to chuckle about by 2032.
To keep cost down expect far fewer SKUs of CPUs & dGPUs to choose from.
For reference PC shipments improved to >275 million worldwide shipments in 2020.
This was before Apple left AMD/Intel with 22.5 million worldwide shipments of Macs for Apple Silicon in Nov 2020.
Leaving 252.5 million Intel/AMD PCs left.
I forecast Expect >80% of all x86 PCs becoming SoC by 2030s.
This would make >202 million PCs with SoC & less than <50.5 million PCs with non-SoCs.
Why? Because Apple's doing it and the PC industry will see it as the easier/cheaper/better/profitable way to make laptops/desktops for the mass market while improving performance per watt, power consumption, battery life & raw performance. This would reduce the complexities of PC's supply chain.
This will negatively impact users who prefer frictionless parts replacements as global units shipped will drop.
I can imagine new Nvida, AMD & Intel CPU & dGPU parts to be released further apart by a few more quarters to even years when volume drops to <20% or <50.5 million.
Nvidia, AMD & Intel SoC parts will increase in volume and may receive far more frequency in newer/faster/better SKUs.
Although AMD/Intel's market share may shrink even further when Windows 11 on ARM will finally address fat binaries & dynamic binary translation as well as those on the Mac.
I could imagine that Android SoC makers Mediatek, Qualcomm, HiSilicon (Huawei) and Samsung will want to expand into the Windows 11 on ARM PC space.
In terms of performance per watt they're between Apple silicon and x86.
Over 1 billion Android smartphones are shipped annually.
So them pursuing the quarter billion PC market would be easy for them to execute. Like Apple most of the R&D cost for PC SoC part has already been paid for by smartphone sales so a little bit more for a SKU specific to PCs wouldn't cost as much as for AMD/Intel.
Conservatively ARM SoC PCs may eat into 20% of 252.5 million PCs.
Leaving Intel/AMD with 202 million PCs for both SoC & non-SoC.
With such a smaller PC market >161.6 million of all x86 PCs will be SoC & <40.4 million of all x86 PCs will be non-SoC.
<40.4 million of the frictionless parts replacement PCs we enjoy today will really be something to chuckle about by 2032.
To keep cost down expect far fewer SKUs of CPUs & dGPUs to choose from.