Watts are a bad metric to use, as they are not always apples to apples between 2 different lights. umol/s is a better way to go to draw comparisons and determine what is the best option. Efficacy is why watts differ.. also, why it can often be cheaper to buy a more expensive light over the 5 years of use. I'll just suggest the total amount of umol/s you want to shoot for, relative to context.
Photoperiods..
Any combination of lights that adds up to 2400umol/s give or take.. preferably err on high side, of course. The more evenly distributed the better. The more efficient the lights, the fewer the watts needed. A mid-to-upper range quality light can probably do this at 2x 480w lights. Cheap lights will need a bit more than 960w, if you want to max out with ambient conditions. Or, mish-mash what equipment you have to reach this 2400umol/s ballpark goal and spread them out as best you can.
If you boost co2, control temperture and RH, then i'd suggest 2x 650w lights with high efficacy. About 3800+ umol/s needed.. and in this case much more probably won't help.. i'd just let the available equipment decide. You'll want a 2.8-2.9umol/J light and those aren't cheap, but in this context you've also spent a lot on ac/heating,dehum, humidifier etc etc.. lol, so don't skimp here!
Autoflowers -- take the umol/s from above over 12 hours and multiple by 66%, since it will run 18hours... if you run longer light cycle (X hours), adjust factor percentage. (12/X * 100 = %) e.g. 20 hours of use would be 60% of the above. it is all relative (amounts to DLI, reference a DLI tabel and you see my suggestions are all in proportion).
If buying new, also prefer a bar/led strip style light. consider DIY. simple angled aluminum does well enough for a heat sink and frame if watts per diode are roughly 0.4w/diode or less. a few hundredths off not being too important.