USB would not be desirable for internal system use, too much overhead. It is well designed for the purpose it has but you wouldn't want it for everything.
But what is "too much overhead" when the transport link gets fast enough? If USB4 ends up with 20 GBit/sec, overhead for anything but SAN shelf backplanes really won't matter.
I actually think I *would* want it for everything. One connector for disks and other peripherals, usable internally and externally. The way they package SSDs now you wouldn't even need to bother with an enclosure.
I think the real problem USB specifically has is a marginal performance history with USB2 devices (high CPU usage, low throughput) and Microsoft's steadfast refusal to allow Windows installs to USB devices, even USB3 (which makes no sense, really, when I can benchmark an ordinary PNY USB3 stick @ 110 MByte/sec read and 60 MByte/sec write).
If Windows could boot off USB3, eSATA would be largely forgotten as faster but with clunky, limited cabling and even SATA as a connector internal standard might get relegated to "enthusiast" boards where some minor performance boost was seen as valuable. M.SATA adoption would end up only in places where extreme miniaturization matters or the same enthusiast crowds.
I know it sounds crazy, but it sure seems a lot less crazy with USB3.1 supposed to hit 10 Gbits.