Why would anyone want to use usb to a dac? If you wanted good sound quality you wouldn’t.
You don’t have to have your data files local to your server/streamer, and if you are using hard disk drives, you definitely don’t want this noise in your audio room.
My dedicated audio room for over a decade, never had a computer in my audio rooms. Even when I had the Auralic Aries many years ago, it pointed to my server with all the disk drives in a different floor of the house.
Do you know how many ms it takes for a request of a disk read of tens of megabytes thru the network, to the server and back, running on a wired 1G internal network? Probably < 200ms. Do you think you will see that kind of a delay? Using ssd/nvme technology will drop that to probably < 50ms.
I used Ethernet to my dac for years until a new streamer came out that provides i2s to the dac, which I bought. The streamer points to an m1 Mac mini which has over 300,000 tracks of music on it and it responds in a blink of an eye when requesting a music file. The network is all wired, Roon installed on the internal ssd, and have a hub which contains a 4TB ssd and a 1TB m.2 drive, all for Roon.
Simple setup, easy to diagnose (which you can’t do with a nucleus/ROCK), and the new Mx apple cpus are the fastest built right now.
USB is the best way to get music to a DAC that does not have a network connection.
For the original question though, as long as both the Mac and Nucleus are on a wired network, and Mac can create an SMB share using whatever is the maximum SMB protocol level that Nucleus supports, it should work just as with any other network storage.
How many of the tweaks/gimmicks have you purchased to try to make usb sound better? Decade ago, I bought into this fallacy (femto clocks, reclockers, cables that separated the data from the power, on and on) and glad I got rid of all that crap and went Ethernet. Also, I knew i2s was even better but I didn’t want a computer in my audio room nor did I want some hack to convert usb to i2s. Now I got a streamer that sounds very good with only 2 outputs, i2s and not usb.
You have to ask yourself why do some of the best dacs/streamers don’t come with usb anymore.
Which ones are those?
I’ll have to ask, too, what are those “best” DACs that do not have USB?
I2S is an absolutely terrible way of moving data between devices, having never been designed for that, and lacking any standard for implementation it is usually a small miracle when it works at all.
FWIW, my new DAC will have i2s, HMDI arc and USB in addition to Coax, optical and AES. So, I can sort of pick whichever digital input is best.
At the moment, I am leaning towards the Roon Nucleus One as an endpoint. Preserve the Mac Mini as the Core and existing library management system, but completely isolate the Mini from the main hifi system by using the One as endpoint. Then, if i ever want to make the One the Core, I am all set.
But, since the One isn’t available at the moment, I’ll have a month or two to just use my Wiim Pro via optical.
Here’s 1, you can read about others yourself:
Berkeley Audio Design Alpha DAC Series 3
You should read up on internals of a cd player and in dacs about how i2s is used. Ethernet is much better than usb (read about that from dcs and in reviews with dacs that provide Ethernet and usb).
So, nothing anywhere near approaching a “best” or even an almost good DAC.
I2S is a perfectly fine way of moving data over carefully designed PCB traces inside a cdevice, where all components are controlled by device’s internal clock. Using things for what they are designed for and all.
When used as an external connector, I2S, apart from allowing for higher bit rates, is about the same as S/PDIF, requires more electrical conversions than USB, brings jitter back into the picture, and creates all kinds of interoperability problems because… oh, right, because it is not a proper inter-device connection, and nobody is going to standardize it, like S/PDIF, AES or USB are standardized.
Ethernet is a perfectly fine way of moving data. If you are moving it over a network. If you are not, it does nothing. From the practical standpoint, on anything that was designed by engineers and not marketers, after data has been extracted from either Ethernet or USB packets, there is no meaningful difference in either data (obviously) or electric noise. No reviewer worth reading had ever either heard or measured any. There’s no reason for it to be there, unlike I2S that can show more jitter, depending on implementation.
So you don’t know how things work. How many tweaks/gimmicks have you bought/implemented to improve the sound quality of usb? Probably none because you think it’s perfect. Well it’s not, it’s probably the worst interface to use for music. Do you know why ifi came out with a usb cable to split off the data from the power? Do you also know why it’s bad to have power coming down the same cable as the music?
Do you know how a i2s cable is built? Do you know how many separate wires there are in a i2s cable? Do you know that it is a better to separate wires in the cable (look at the ifi cable again). So if i2s is already implemented in audio equipment, why would you want to go thru many conversions to get to the desired destination?
Why are there reclockers for usb connections to the dac? Why is there equipment to remove jitter from usb? You have none of these gimmicks/tweaks for i2s or Ethernet.
I have a top quality dac that has both Ethernet and i2s along with usb/toslink/aes. Ethernet was better than usb and i2s is better sounding than Ethernet.
I’d say do the test yourself but you have no such dac to test these interfaces, you just have inclinations on how things will sound
Because fools are easily separated from their money
I agree. I tried many things years ago to try to get usb to sound good, even the so called audio usb port out of a dedicated streamer. I didn’t go all out for the off-ramp or some of the other higher priced items. I was already into a couple thousand $$ of tweaks/gimmicks for usb before I decided to get a better dac that had Ethernet and i2s. I could have bought a much cheaper dac that 99% of the audio people buy with usb/todlink/spdif connections, but I also don’t want a computer in my room, 1 with spinning disks back then nor a computer with a fan, I want it dead quiet. Ethernet allowed better sq and be able to keep my computer in another room. I could have purchased a usb to i2s piece of equipment but I wanted to stay away from usb. A new streamer came out with Ethernet in and i2s out, which perfectly compliments my dac
I actually do. And no, I’ve never replaced fuses with “audiophile” versions, never used isolation platforms under solid state electronics, and never put any magic crystals on top either. Because I know how electricity works.
And that would be because?
Because iFi can not compete on the quality of their products (their engineering is only concerned with lowering costs) but can well compete on selling a cool, if nonsensical, story.
Only if you do not know how to design electronics. There is no “music” going down USB cable (or Ethernet, or I2S) anyway.
Because, if you had some idea, you would have noticed that I2S that is inside the equipment, where it belongs, is designed for being sent over (hopefully, properly designed) PCB traces of specific length and shape. I2S that is sent between devices, using some cable repurposed from an entirely different connection, needs to go through more electrical conversions.
Because there are enough clueless people who buy them There are “audiophile” fuses, too. Does not mean they have any effect either. And in any case, USB does not have “jitter.” I2S does, but not USB.
Seriously? You’ve never heard of “audiophile” Ethernet switches and regenerators? Don’t you know that Ethernet sounds terrible, terrible, I tell you, unless you run it through an expensive reclocker?! 
Well, and finally we have it, the ultimate audiophile argument. “I’ve spent more money on the equipment, so of course I know better!”
You should just learn why I2S interface was popularized by a certain 0-engineering company in the first place.
I2S is commonly used from a controller chip (inputs: I2C, USB, coaxial, optical) to a DAC chip using I2C topology. Such connections are PCB mounted with close proximity. This type of connection is synchronous and jitter is determined by the controller chip with precision quartz oscillators, eg. x44.1 or x48 clock samples.
If I2C is used as external connection, this synchronous connection has to depends on the source, e.g, a CD transport or streamers. When a clock signal travels down a long cable, due to capacitances and inductances it will cause significant jitter. Provided, the I2C clock is re-clocked again before going to the DAC, I wouldn’t recommended this type of connection.
For USB, it is asynchronous so jitter is depends on the controller to the DAC chip. The only problem with USB is noise from the source. There are many to get around it, such as USB isolation and USB Regen.
I don’t know where you got the pros and cons from, but I think they need some work.
- Jitter is a problem for I2S-on-a-cable, which is synchronous, and not for USB, which is, in most DACs, async.
- While USB in general supports data integrity, it is not used in async mode as far as I know.
- Regarding complexity, UAC 2.0 is supported by virtually all DACs and OSs, so it’s not an issue. On the other hand, there is no standard for ‘external’ I2S.
- Regarding latency, I2S is indeed designed for audio transfer, but between ICs on an interna bus. It’s not designed for any types of cables, long or short. Once you put it on a cable, you can introduce jitter (and of course noise), and in order to eliminate it at the DAC, you need buffers, PLL etc., which add latency.
I2S for DACs is just a marketing gimmick that doesn’t solve any problems with any of the standardized audio interfaces.
I suspected it was AI, but you should have disclosed that.
You’re confusing a lot of interfaces here that don’t relate to each other.
To answer the question in the subject: It depends on the DAC. Many DACs have wonderful USB interfaces these days. A poor USB interface should not be used. A DAC with I2S + USB, probably use the I2S as that’s probably where the manufacturer put their money where the USB is there check a box on a glossy press pamphlet.
Also, what are you selling? 
Your post reads like the garbage at the top of a “white paper” masquerading as pure advertising fluff trying to justify I2S from a manufacturer who, as you put it, “don’t know how things work”.
Oh, well, there you go. USB was garbage a decade ago. It’s no longer garbage and is very competent in a lot of implementations. It’s amazing what quick gimmick fixes audio companies will sell you until they figure out the right way to put an interface on a circuit board. You should try it again. In fact, you might even find that the higher bit rates it provides can be very pleasing.
Also, what streamer are you using? I’d take a guess based upon this post and description. There are only a limited number companies who are still married to I2S.
Yet another example of AI being very artificial, but lacking any sign of intelligence 
Don’t blame AI, blame the training material (most likely the web).
Sure, but people seem to think that there is some actual “intelligence” to it. But there is none, it just regurgitates whatever nonsense it was fed originally.
If usb was better today, why do you still have these devices for sale? It’s only $5000:
As for usb, I have tried usb lately. I no longer have my old AQ Diamond usb cable, but I have an AQ middle of the road usb cable. I prefer to use another interface than usb.
The only drawback to using i2s is the requirement for a short cable. I’m using a 15” i2s cable that cost $1000.
More and more audiophile dacs are including i2s.