I don't recall any functional issue using avb under Windows except that it means you can't use network control unless you also buy the avb switch (i.e. without a switch, the extra device has to use the network port on the master device). I don't think avb is relevant unless you have multiple devices is it?
AVB doesn't really exist on windows IIRC. Hono had a virtual ASIO device that utilized an RTX operating system + i210 nic to talk to AVB networks but that's about it. They don't sell it anymore either which raises questions about it's current compatibility if I were to say, find a license or figure it out.
Obviously you can just use the USB ASIO drivers with an AVB Ultralite, and get audio out of a windows PC into an AVB network but you can't do 'no usb' IP only AVB, which is extremely compelling.
Also yeah, the switches sold by MOTU are way overpriced. Turns out Netgear GS724Tv3+ switches have great AVB support if you buy an EAV license. The switch + license comes out cheaper than the other options out there.
AVB, and IP audio along the lines of AES67 is pretty rad because of it's flexibility and expandability. MUCH less hassle changing or expanding setups, and since the main target, at least for me is higher channel count scenarios, cutting down on massive snakes and allowing D/A gear to live closer to amplification, without having computers nearby is obviously ideal.
Furthermore, jitter and other possible digital problems like 'incorrect' usb frames, don't impact proper IP audio, unlike USB Async communication.
USB Async might 'know' a frame / sample is wrong but it can't do anything about it beyond making a decision about what to send the D/A chip.
With IP audio? Say a packet doesn't look right, well the receiving device can actually ask for it again, and because of the way the timestamping works, it can actually take the re-sent, 'correct' packet, and slot it in the buffer where it should be BEFORE it has to send it to the D/A chip.