Does that mean there's no possibility of having the data write to an microsd card?
It doesn't mean that at all. The problem is that writing to the SD card is not interrupt driven. Most writes actually write to a buffer, but, when the buffer gets full, the buffer is actually written to the file. During that time, which can be lengthy, you can not be sampling audio data.
That, of course, will be a problem.
Then, there is the issue of actually streaming the audio data from the SD card over ethernet. Unless the other end is buffering the data (which it probably is), don't expect that data to be useful for generating sound. Even if it is, there will still be a long lag before enough data gets transferred to be useful.