It may use compressed air to nitro and partially carb the beer. That would only work if it carbs on demand, or it would oxygenate your beer. For that price there's no way it scavenges carbon dioxide from fermentation and liquefies it (EDIT: if it didn't liquify it, it would need a big gas bag/balloon and that would look exceptionally daft). The thing would need to be serviced like a car. I'm going to guess that it's just pressure-fermentation with a solenoid valve that closes after a period of time, and/or a back pressure regulator (not the same as a relief valve). Just like those horrible little plastic barrel things you could get from kmart.
It is completely possible to build a fully automated brewery. There's no reason why the quality wouldn't be as good as anything else.
From an engineering standpoint it's not difficult.
What makes me highly skeptical of any of these attempts at full automation is that, well, it isn't full automation. The easiest way to reduce the control system complexity is to reduce the process complexity. This is what they've done. "Let's just ferment goop that's way easier than trying to control all-grain wort production, wort cooling, CIP etc"
When I see a mini automatic valve matrix, CIP tanks, pumps, refrigeration (either DX or reticulated water/glycol), mash rakes, grain silos and grain transport and countless other processes/items that are proven again and again to be necessary to make good beer, then shall I say that a truly automated beer machine has been achieved. But until then, colour me skeptical.
