There should be discussion of this in some of the FSD threads, but if anyone here feels competent at assessing Qualcomm's release today of Snapdragon Ride - a hardware platform for automated driving - such discussion will be encouraged. TechCrunch and Reuters, among others, have stories on the development.
Qualcomm's particular strengths here would be very low power computing, AI, very fast and reliable wireless communication and integrating various auto sub-systems seamlessly and with great efficiency. They have been working on these things for decades and tend to be able to attract top talent. Judging solely from the Reuters article, it doesn't sound like their self-driving program is very advanced, being mostly a bunch of sub-routines for lane-keeping, object detection and automatic emergency braking, probably image recognition, and lots of other things. But we know FSD is a difficult problem and their routines probably don't get them close at all to FSD (the hardest part is the last 10%).
I think it's probably a solution for automakers that are struggling to integrate over-the-air software upgrades, advanced driver assistance features, etc. Any FSD that came out of this program would likely look very different from Tesla's in that it might rely on cars in the vicinity communicating their intentions to each other (and perhaps specific data about other cars in the vicinity) wirelessly from car to car or over the 5G networks. Hence, it would probably only be fully functional with roadside microcells (which are becoming silly cheap). Obviously, this would, by necessity, be a hybrid system that used current driver-assist functions in conjunction with a wireless network of connected cars sharing information.
This is just my speculation and obviously, any such solution would require a common standard, wirelessly equipped cars and governmental endorsement so it would be the likely result a decade or so down the road and would only be absolutely necessary if vision, lidar and radar-based FSD systems cannot be developed to a high enough standard to be fully autonomous. Qualcomm might be a likely candidate for leading the standardization effort. In reality, the industry would form a standards-setting body and all would participate. It could be based upon the cellular standardization process although it would likely be a process much improved and better designed with the experience gained in cellular. Qualcomm originally came to the cellular standards game late and was only successful because their technology was light-years more advanced than the nearest competing standard. They had to fight big monied interests who gamed the system to extend the value of their new (but already outdated) wireless technologies another 5-10 years. It was messy and inefficient due to participants protecting their own ability to monetize.
Of course, if vision, lidar and/or radar based systems are developed to a high enough standard, none of this would be necessary. On the other hand, even with complete FSD, wireless communication could be used to improve and enhance the safety, speed and efficiency of traffic flow. Actually, I believe it's inevitable that FSD cars will be wirelessly connected within 10-20 years even if they are fully cabable of getting by on their own.
Qualcomm would have bought out NXPI in 2018 had the Chinese government not witheld regulatory approval as part of a "tit for a tat" spat with the US government. It would have actually been more of a merger as NXPI had a lot of synergies with QCOM with their considerable advanced automotive systems expertise while they were lacking in digital wireless and low power processor tech that QCOM possesed. I believe QCOM and NXPI are likely still on very friendly terms and have probably been sharing information and working closely together on automotive tech since the merger/buyout was squashed. NXPI is probably a supplier of high power silicon to Tesla but I'm sure someone here would know more about that than I do.