Just a few clarifications...
Even if the design changed to a different architecture it will still have to be fabricated somewhere. Due to the costs involved it will be much cheaper to use a foundry than to build your own fabrication facilities. I do not see how RISC-V will change the current dynamic towards foundries being more relevant. Also I think ARM pretty much has the smartphone market locked in by this point. RISC-V would have to come through in things like cheap IoT devices and the like if it is to have a chance I think.
But Intel already tried selling foundry services and failed. There is no evidence they will do better this time. I think this is just talk by Intel so they can get US and European funds to increase capacity.
I don't think it will change the dynamic. I only think there might be some opportunity for other foundries exposed by any shift away from ARM. As I understand, some of TSMC's current strength is that they have the best libraries for low power ARM designs. So I think the last Snapdragon (888?) was dual sourced between TSMC and Samsung, but TSMC was still running cooler/consuming less power at the same nominal process size.
The failure of the initial push as I understand and the previous article pointed out was mainly because Intel's foundries required Intel tools to build. They were not industry standard, so people didn't want to learn to use them. Some reports also said they were not really suitable for things like low power SoCs, and geared towards complex CPUs. According to this "2.0" strategy, they will adopt more industry standard layout tools like Cadence. (Again, basically the limits of my technical knowledge here)
Overall, I do agree the strategy is driven a lot by the carrots being dangled by western governments. You see the announcement today for their new facility in Ohio and there were strong hints to the need for more incentives.
You just contradicted yourself. Apple designs their own ARM processors. Also Qualcomm does have lots of competition besides Samsung with Exynos among Android CPU vendors. MediaTek for example is a major competitor which even has their own modem designs.
Just to clarify what I meant, Apple is obviously a special case. They are not selling their designs and 100% of the production is for their own platform. Therefore it is not competition in the CPU space, at least not in the way Qualcomm vs. Mediatek vs. Unisoc, etc. is.
Qualcomm does have competition, but not really in the high end/high margin space, at least not much serious competition in the last 5 years. Intel as we know loves high margins which is how they got sucked away from low power optimization in the first place.
In the low to mid range markets, you don't need the cutting edge processes to extract maximum performance. I think Unisoc is still using 14nm.
As you noted, most of the surviving SoC vendors have integrated modems, that includes Samsung. Intel's failure in this respect probably would have doomed it's effort even with ARM/XScale, which was my point.
As an aside, HiSilicon was probably Qualcomm's big threat. It had the most capable modem at the time and comparable performance. Was Huawei willing to sell to outside vendors? There were some beliefs that it would because it would help their more upstream network equipment sales.
Intel is not this all conquering company. They routinely fail at a lot of things. This is just yet another one of them.
Lol, I owned a K6 remember? That was followed by an Athlon. I also owned an iMac G4 lampshade style. I’m not Intel’s fan at all, but I wouldn’t count them out
