Hydroponics doesn't scale in the way that traditional "stick a seed in the ground and leave for months at a time" scales, regardless of if you are doing it organic or not. This is also not the first "robot that will solve" agriculture. There is a mentality in silicon valley that they are smarter than everyone and just no one else has thought of these things. But this idea in principal has been around forever, and I even saw some demos at experimental stations I was working at back in the early Obama years. Pretty sure none of that went anywhere, but then again, robotics have gotten cheaper, etc.

The two best tech's I've seen in this particular space were 1, basically a small hydraulic ram that beat the ever living piss out of small seedlings, and two, a high powered laser that smote small weed seedlings with the power of science. Either way, the places I've seen this work, its only on demonstration scale; very contrived circumstances; and always on extremely young weed seedlings. There also almost always seems to be a bit of cheating around the particular weed species. Consider the computer vision aspect: you need to destroy the seedling when its very young; many crops have weeds that are closely related and indistinguishable at that stage. So you'll see demonstrations where its a dicot weed species in a grass species crop or vise-servus (or like, cheat grass in rapeseed, or some brassica in wheat).

But yeah. Inevitably it ends up being other issues that derail this. Small robots need to be kept track of, powered, etc. They get mud in their tracks. Birds shit on them. Crackheads steal them all and sells them for parts. They fall over sideways. And like, yeah you could build them bigger and heavier, but then you've just reinvented the tractor.