There are various theories about plasticity and things happening on a different timeline. My read on that exclusion criteria for this study is that they're a VC-backed for-profit company trying to maximize the efficiency they can demonstrate to make it easier to get to market.
I'd say I'm agnostic on the first issue, but agree that the second is a significant concern. Double blind RCTs are the gold standard; double blind RCTs are what UMich is doing, and this makes me a lot less suspicious of the UMich data despite smaller sample sizes. (I think UMich, so far, has shown somewhat less dramatic gains than what Neuromod is claiming? If so that all lines up neatly).
I'm agnostic on the first issue because we don't really know enough yet to say with any certainty at what point plastic changes happen in various brain regions, and I don't think it's impossible that bimodal stimulation could work better in general on people who have had tinnitus for X time frame or less. I am not privy to Neuromod's initial data, and they may have some reason to think so. On the other hand, UMich doesn't seem to care, because they're more interested in finding out as much as they can about how this stuff works and who it works on, than they are with the typical insane VC rush-to-market. So, we can say all the usual things about Neuromod that can be said about VC-backed capitalist endeavors, but I personally don't resent them for their existence; when it comes to capitalistic silliness, I hate the game a lot more than any specific player (excluding really dramatic examples of bad-faith actors like Dow Chemical or the petrochem industry).