If I were them I wouldn’t. The world is littered with boutique DACs made by companies who area of expertise was definately not designing and implementing a good DAC - some get the digital side right and fail on the analog side, or get the analog side really good (because they are historically experts in analog audio gear) but fail badly with the digital side and USB noise rejection etc poor firmware/drivers etc.
Ive seen many companies loose badly with DACs in the pro-audio / DJ space because they just didnt have access to the entire skill set and resource to do a good enough job. Granted, now its much easier now with class compliant drivers etc.
Also I may out them in a awkward position with partners. Right now there is no competion between Roon and their partners. Roon focus on core and software, partner focus on audio and digital transport hardware. Add in Roon operating in that space and I can imagine it upsetting the balance, possibly to everyone’s long term detriment Partners will be less inclined to be as open about the internals of their product for competitve fears (Ive seen this in other software/hardware segments, though the company I was working for at the time was and is big enough not to care who they trample on).
I really think the whole ecosystem will remain in a better place if each focus on what they are best at and keep developing good partnering relationships.
if Roon were to step into this space, they would need to do a really good job at the price point for the sake of brand reputation. And in keeping with their brand, it probably wouldnt be a mid range 500ukp or so device - probably 800-1k at least and that would put up against some really good and highly experienced competition.
Maybe they could do a network endpoint DAC driver box - that at least would be a reasonable companion to Roon, but it would still step on some toes. It may also mean them perhaps crossing a line that I dont think they have ever crossed - promoting some idea of better sound quality through subjectively debatable elements in order to position against existing alternatives. If they ignore this aspect then then maybe only a very low price point could be met which again probably wouldnt be worth their while. currently in terms of crossing the line in the subjective world, they leave the options open to the user, document the method but don’t really take a position. With and endpoint they probably have to cross the line.