Also you haven't addressed stacking, heat retention or for that matter nighttime light.
Break even would take 50% efficient solar collectors and 90% efficient lights including transmission and conversion losses. Clearly, we are nowhere near that point.
PS: The only reasonable option would be solar collectors that focus on different parts of the spectrum and let light through to the plants under them.
And again how do you stack plants? How do you keep them warm enough? How do you grow them during the night?
You can play with the arrangements of plants in a greenhouse just as easily as a factory. EX: Stacking plants using clear trays and nutrient bath allows for using more light in the same space.
Anyway, the reason why this 'works' is crazy subsidies in Japan which actually has quite a bit of farm land, plus a market for ultra premium lettuce for decorative reasons.
And no you can't stack in a greenhouse for the same reason that there are hardly any plants growing on the jungle floor.
This sort of thing is already being done in London and Sweden. It's the level of automation that is making this news. Of course it starts at the high end but eventually all food will be grown like this just because it'll be the most efficient overall.
You focus entirely on the efficiency of photosynthesis in plants when there are so many other variables as for why this is a good idea.
Shipping bulk rice from US to China is cheaper in terms of energy than shipping electricity required to grow that same Rice.
PS: Another sanity check. The US has 922 Million Acres of farmland. At 1kw/m * 8h / day that's vastly more than the world total energy usage. You can't even replace that with fusion power as the waste energy would directly produce significant global worming. (4046 sq M / Acre * 8kwh/day = 32 Megawatt hours per Acre * .92 billion = 30,000 Twh per day, world electricity is under 30,000Twh per year. ) https://en.wikipedia.org/wiki/World_energy_consumption#/medi...
Also you have to take into account increased yields, faster grow times, lower fertiliser, pesticide, herbicide and tractor fuel costs.
Your sanity check is way off. Solar panels only get about 22W/m^2 though out the year on average and they're more efficient than plants. Second the maximum plants can use is about 100W/m^2 and that is at maximum midday strength. Now lets take into account varying intensity of sunlight, growing seasons, leaf orientation, gaps between plants and other factors and your energy requirements are much lower.
Research is still being done but most plants grow very well with the full red spectrum and 10-20% of the blue spectrum. This vastly reduces the energy required.
If you really want more farmland and have a lot of energy there are simply much better options, desalination and irrigation of west Texas for example.
PS: You can bump plant efficiency from 10 to 20% from using a narrow frequency of light, but that's a fairly low cap. And 10%(plant) > {20%(panel) * 20%(plant) =} 4%, which is why I said you need to have 50% panels to 'break even'.
Second, photosynthesis is actually fairly efficient, the issue is plants are living organisms and we only care about energy stored not energy used. Net result, plants still need about 45% of sunlight even if you stick with the 'perfect' bands.