Hero, in this case the issue is a bit different. If you can emit discrete spectrum light that matches the absorbtion peaks of the plants you want to grow than you can be more efficient with led. The phosphors of the fluorescent tube makes it impossible to have sharp spectrum bands.
There is just a whole bunch of ifs attached to this statement.
- You need to get 660 and 460 band covered. 460 is possible to find in high power, 660 is not. All the red leds are far past the peaks.
- With a 80-90% efficient custom made smps, led would be a nice setup, specialy if it is a complement to natural light or a source of continuous spectrum light to cover lesser light related phenomenon in plants. You are not interested in lumens but by sum over the spectrum of the product of emitted power spectrum by the absorbtion function of plants.
- a good led based system is going to be much much more expensive. For 100W output, even supposing efficiency equal, the tube setup is going to cost less than £50, a led setup is going to cost 150£ at least.