[Wytwwww]

IMHO it might be the complete opposite. Margins are low and and off the shelf ARM designs are very cheap so what's the point of making very risky investments when you can already get good enough off the shelf designs from ARM?

If the licensing fee is only ~1-2% who can justify this besides Qualcomm, Apple, Nvidia, etc.? And Apple and Qualcomm only did this because they wanted to (and could afford to) design faster chips than ARM.

[snvzz]

There's the option of licensing a RISC-V microarchitecture, rather than an ARM one.

It has the advantage that RISC-V is an open standard, and there are many companies licenses for their own RISC-V microarchitectures, rather than being stuck with licensing ARM cores from ARM.

[Wytwwww]

From whom? I'm not sure why would anyone who can design high-end RISC-V cores would license them to someone else instead of making the chips themselves. As ARM has demonstrated that's not a very good business model (they are barely making any money compared to some of their customers).

[snvzz]

>From whom?

From any of the IP vendors offering them.

e.g. Tenstorrent has Ascalon (already licensed by LG) and Alastor, SiFive has P670 and P870, Akeana has the 5000 series.

[Wytwwww]

I just don't understand why would anyone who developed a competitive mid to high-end core would want to license it to anyone? What would they have to gain by doing that instead of making them themselves?

Of course there are cases where the CPU is pretty much just a "cost centre" (embedded, industrial products and such) but I don't see that working in the PC/mobile/(other general purpose device) market.

Doesn't really seem like a sustainable business model long-term i.e. you won't be able to generate enough revenue to compete with companies designing chips in house.