No, that really is impossible. Mother Nature's little growers are optimized for the amount of sunlight they receive on flat ground, and by trying to stack farms on top of each other you utterly inevitably mathematically reduce the amount of sunlight each floor can get to substantially less than that, because the floors occlude each other, except the top floor. To the extent that people fail to take that into account they really are failing at physics badly and Monbiot is correct to criticize them.
I think under the right circumstances and with plausible technological development this can make sense, but he's absolutely correct that doing it with current technology is completely infeasible and useless. My position is actually not what he is criticizing, and I'm still open to the possibility that while the future may look different, the actual best solution is something not yet evident. We may also simply reduce the costs of transportation such that growing something 1000 miles away simply isn't that big a deal ecologically. (We really, really, really need cheap energy, cheap in all senses of the term, ecologically as well as economically. And there are feasible possibilities.)
And sub-10-story warehouses only magnifies the problem-- then you need to buy even more city blocks.
Purchasing land 30 miles outside of a city falls well within the confines of urban agriculture, especially since indoor ag means a year-long growing cycle. This means that Detroit or Iceland don't need to rely on South American growers for certain seasonal items.
The small shipping distance necessary for putting these in New Jersey, greater Chicagoland, etc., there's no argument against that, but yet, people keep citing downtown Manhattan as impractical for farming. Of course it's impractical for farming, it's damn near impractical for living.
Is that actually true? Crops are grown in a pretty wide range of locations, and sometimes the same crops are grown in different locations with quite different insolation. Surely there must be at least some set of [location,crop] pairs where the crop needs less than the full amount of sunlight available in that location?
I also find it hard to believe that you could find 10 layers' worth of crops this would work with. Two or three, maybe, but I would be astonished by more than that.
As for lighting, the trick is to figure out how to use natural lighting for a majority of the input. Obviously, again, this is why 100-story skyscrapers aren't practical, but sub-10 story warehouses could be. Although some are investigating architectural ideas like mirrors and other hacks.