[monocasa]

I'm running off of the assumption that Thunderbolt continues to use a different physical layer (which it pretty much has to to go faster than USB 3.1), and is piggy backing on USB Type-C's "Alternate Mode" which lets you negotiate different physical layers that you have a transceiver for.

EDIT: Looks like my assumption is correct:

http://en.wikipedia.org/wiki/USB_Type-C

[bryanlarsen]

Sure, but why wouldn't the TB3 transceivers be designed to support USB 3.1 as well? Those transceivers have a wide range of dynamic adaptation to support high speed transmission, so it doesn't seem like much of a stretch to adapt to USB 3.1 signalling levels & protocol.

[monocasa]

Because even above the physical layer, the two protocols are very, very different to the point that you're nearly doubling the amount of work that you have to do. USB at it's core is a host polled interface that looks like a network, and Thunderbolt is at it's core a multimaster RDMA interface. You'd basically be designing two different devices.

[bryanlarsen]

Sure, but USB is a very cheap protocol client-side. I fail to believe that requiring it in all TB3 clients would be prohibitive.