"As high as possible efficiency" in terms of ...what, exactly? Sure isn't going to be electrical cost if you are dumping in enough heat that you need artificial cooling as well as the power for the lights. Supplemental lighting in a greenhouse at least has the potential to make use of some of that heat in a positive way, but heating the greenhouse overnight is often enough of a fiscal drain to make the plant products excessively expensive anyway (depending on the exterior climate, of course - in milder "but still too cold for basil" climates less supplemental heat is required than where I'm at and base my greenhouse cost assumptions on.)
Mean midday solar radiation in Thailand in Springtime averages 800W/sq m and HPS lights are far from 100% efficient at converting electricity into light, so you ought to be safe if you can dispose of the heat. 800 "PAR Watts" (photosynthetically active radiation) per square meter should be "safe" if you can control the temperature, but might take 2500 watts of electrical input for HPS lights (and getting rid of 1700 watts of heat) based on some quick and shoddy research. MH might get you down under 2300 watts input for the same light to plants. But see below...
Revising (slightly) the shoddiness level of the (still pretty shoddy) research, perhaps a factor of 0.368 should be applied to the solar irradiance to get it's PAR fraction, leaving 295 PAR watts/sq m and revising the equivalent MH wattage down to ~850 and HPS to 950-ish.