One question I'd ask in return is, do you expect ultralow-field NMR to replace high-field magnets? You'd likely agree that this is impossible, given how dependent other fields of research are on high-field NMR. So it's not plausible for universities or other institutions to get rid of their high-field setups and replace them with ultralow-field NMR.
Once we accept that, then the question becomes: what is the benefit of adding ultralow-field setups to an existing high-field NMR facility? Given that the low-field spectra are difficult to interpret (after all we're talking about magnetic fields several orders of magnitude smaller than what we're used to, and possibly with multiple nuclei overlapping with each other), I don't see why people would be in a rush to get their hands on low-field setups, when they can get their 500 MHz spectra and go off happy. On top of that, there's an obvious barrier to entry in terms of technical knowledge required to build and operate such a spectrometer.
I certainly don't mean to discredit the existing research that has been done on it, of course. It is not bad, or useless, to perform research on this. However, realistically speaking, and as of the time of writing, I don't see the incentive for this to be "the next big thing"... yet.
The current trend in "lower"-field NMR is probably benchtop NMR spectroscopy, and (apart from arguments like portability) that's almost certainly only because it still gives you pretty decent spectra (e.g.., it's suitable for undergraduate chemists to look at and apply the $n+1$ rule, etc.). If benchtop spectrometers were 1 kHz I doubt anybody would be buying them.