CO2 assimilation rate depends on concentration and species. For example, at 400 ppm, butterhead lettuce would assimilate about 0.38 lb of CO2 per ft^2 per hr. Whether greenhouse or bomb shelter, CO2 could be replenished through ventilation, infiltration (i.e. unintentional ventilation) or supplemented (e.g. from tanks). There are benefits to increasing CO2 concentration above ambient, mainly that the plants need much less light to achieve the same growth. Experiments at Cornell have shown lettuce needs about half the light if you increase CO2 to about 1600 ppm.

It's pretty hard to get good numbers on this from manufacturers, because almost everyone gives specs in human vision units (lumens, lux, foot candles, etc) if they give any information at all. For growing plants, photosynthetically active radiation (PAR) units are required to make sensible comparison. PAR consists of wavelengths between 400 and 700 nm. Conversion between the measurement systems is not trivial and can only be done accurately if you know the power of each wavelength in the spectrum.

It's also very important to include the ballast power for HPS. When you see something like 600 W or 1000 W HPS, that does not include the ballast. Same goes for including the driver power for LEDs.

A typical efficacy for HPS is 3 mol (PAR)/kWh. We are starting to see LEDs with efficacies above that. But there are other considerations. HPS is basically a point source where most LED fixtures are rectangular panels or strips. This is a concern in a greenhouse because you want the luminaires to block as little of the sun as possible, and current LED fixtures shade a lot more than an HPS system of equivalent light output. Putting strip LEDs directly under the structural members might be viable, but then they would need to be more powerful and/or have lower ceilings. Lower ceilings can cause issues with ventilation and temperature control. LED fixtures are several times more expensive than comparable HPS systems.

In either case it's pretty difficult to get a uniform spread of light intensity over the whole canopy. Specialized software and sometimes custom reflectors are needed to do this optimally.

Upgrading from fluorescent to HPS is a no-brainer. But HPS to LED, the picture isn't so clear (in agriculture). I think LEDs look very promising in the long run, but they aren't quite there yet. There's a reason why nearly all big commercial greenhouses use HPS.

We've barely scratched the surface in researching this area, especially the subject of what effect different wavelengths have on plant growth. At Cornell we have some upcoming experiments that will be studying the growth of vegetables under several types of LED and HPS lamps.

At Cornell we've done lots of research in the area of Controlled Environment Agriculture (CEA) which includes hydroponics technology. There's a lot of bad information on the internet (and in this topic's comments). Just a few points:

Re: LEDs being more efficacious than the sun, it's not so simple because you don't pay for direct solar, where you do pay for electricity used to produce supplemental light (whether in a greenhouse or a bomb shelter).

Even in a relatively cloudy climate like Ithaca, NY, lettuce can be grown year-round with about 70% of the required light coming from the sun (free) and the other 30% from supplemental light. In the summer you often have to use operable shades or whitewash to avoid tip burn because you are getting more sunlight than the plants can handle. These results have been verified in actual operating greenhouses. Some of our recent research is looking at the HVAC side of the energy picture. A paper due to be published next year shows that it takes about 3 times the energy (lighting + heating + cooling) to grow an equivalent yield of lettuce in a warehouse-type structure than it does to grow in a greenhouse, as long as it is properly designed.

Sunnier climates, such as Long Island, can get about 85% of the required light for free from the sun.

Growing indoors with 100% supplemental light has not only an enormous cost for the light electricity, but also a significant increased cooling load. You also have to take into account the larger peak load in kW as that affects your demand charge from the utility (a big number at these power levels). Even with a very efficacious LED, nearly all of that light energy transforms into heat which must be dumped out of the building.

Solar panels are great in general, but substituting them for free direct sunlight isn't even close to economically feasible because of current efficiency.

Beware of specs you see on the internet. A big red flag is when lighting manufacturers talk about foot-candles, lux, or lumens. All of these units are normalized with respect to human perception of color, and are therefore meaningless when it comes to plants. The proper units deal with PAR (photosynthetically active radiation), measured in mols/m^2/s or something similar. There is no simple conversion between the two measurement systems, because you need to know the exact spectral distribution and that varies by light source. Of course none of this stops manufacturers from selling lots of these fixtures to uninformed folks.

Most large commercial growers use high pressure sodium (HPS) for supplemental lighting in a greenhouse. We are doing some rigorous experiments to compare HPS standard luminaires with some LED fixtures. But so far the installation cost premium for LEDs is typically prohibitive.

Indoor farms and vertical farms look neat, but we've never seen a peer-reviewed paper demonstrating how they could be more energy efficient than a well-designed greenhouse. And we've never seen a indoor/vertical operation be profitable without subsidies, huge markups on product, or both. I'd love to hear about a counter-example.