While I agree with you wholeheartedly about ThunderBolt's low latency and transport of PCIe beyond the confines of the locality of the motherboard, and that there are some applications where undoubtedly this is ideal, such as external GPUs and some high end audio processing, I really don't see TB as the best interface going forward, and I really don't see the industry jumping on TB, mostly because of political reasons rather than technical merit.
The biggest reason I see TB as having any traction at all is Apple's forcing it down the throats of their users whether they wanted it or not.
TB cables have chips in them to facilitate proper operation and actually need the chips to attain the touted bandwidth numbers, plus a two second search of google shows these cables tend to cost well over $10, which to a lot of the world market is an issue and limits adoption. Are we really concerned about host/target controllers at this point since both the host and target actually end up having some sort of controller anyway? Most smart phones have processing power to spare, which isn't to say I agree with wasting CPU cycles, but I'm also not a proponent of requiring expensive "smart" cables either. Most consumers aren't looking for expensive solutions. This added cost works against TB and keeps it more niche.
You bring up rumors, which I'm not against, but if you want to talk rumors, USB 4 is rumored to have 100 Gb/s bandwidth, and final specs may be announced as soon as 2020. I have run across no rumors about TB 4 and have no idea when the specs are supposed to be announced, if they even have enough market share left to come into fruition.
I think I can understand if Intel is leery of certifying TB equipment for the TR4 platform, as on a lot of Intel equipment, TB is routed through the PCH controller and it's 4 DMI lanes, and as such may show an unfavorable situation on Intel equipment, whereas on TR4, the PCIe lanes route straight from the processor, eliminating that potential bottleneck. I still see even this as niche outside of review sites and select few parties that have an interest in the drama it promotes. How many real world situations are pushing a 40 Gb/s interface to it's limits on a regular basis!
Companies have worked around USB's latency issues since it's inception, and can continue to do so. It isn't as though the multi-track recording industry with effects added began at the onset of TB in the marketplace. Companies that choose to forgo the wider adoption of the USB interface for the TB interface can continue to do so. There should always be plenty of ways to get the same job done.
I can also understand if the market for equipment that truly needs what TB brings to the table is so low that the cost to pursue it for AMD, Gigabyte, ASUS, etc., is higher than what they see it bringing in sales.
I see newer USB revisions as the market's interface going forward as history shows it's been well accepted, despite it's issues. Whether Intel capitulates the need to jump through the remainder of their hoops or not for certification I'm unwilling to wager on. They may be a company made of smart people, but they've made their fare share of silly decisions, and I'm not going to bet that hard heads suddenly become soft, just to save an interface.
The current situation seems unfortunate, as I like the idea of pairing external GPU solutions to small form factor computing solutions. Here's to hoping that the more open USB standard grows to include the functionality of TB, rather than as it is now, the other way around.