RME actually blames Intel marking some Thunderbolt chips as EOL. And that's maybe plausible if their hardware doesn't look like a PCI device on the other side and they're using Thunderbolt in some other way. It's not really the way Intel expected Thunderbolt to be used, I don't think, which is why they probably didn't expect EOLing a chip that's designed for use on PC motherboards to break a bunch of device vendors, but that's rather beside the point.
The AKM chip factory that burned was five years ago, and they built A/D and D/A converters. That's clearly not the supply chain issue causing, for example, the LP32 to be unavailable, given that it has no analog audio at all.
The ASML fire mentioned elsewhere in the thread is a red herring. Their building had only minor damage. The plant was partially back online within days, and fully up and running by about four months later.
If they're having to try to buy NOS Intel Thunderbolt controller chips from anybody who still has them lying around because you don't want to do a redesign with a current-generation chip, that could explain why they're only making the most popular products and leaving those of us who are waiting for other products high and dry. IMO, they'd be better off ditching the original version of Thunderbolt and moving to Thunderbolt 4 or 5 (perhaps configured to run at Thunderbolt 3 speeds for broader compatibility).
Then again, they'd arguably be even better off putting a five-port AVB switch inside the device, making their device pure AVB, and making the Thunderbolt port just be an AVB-compatible Broadcom chip hard-wired to the switch. That way, their Thunderbolt hardware is just a bog-standard off-the-shelf retimer connected to a single-lane PCIe bus connected directly to the Broadcom chip with a 10 cent 8-bit microcontroller to handle the USB-C configuration pin handshake, similar to what Apple did in their Ethernet adapter, but maybe with a newer 2.5 gigabit chip.
.)