Current LEDs are around 40-50% efficient. So 45% efficient LED * 25% efficient solar = about 12% total efficiency.
[1] A. De Vos & H. Pauwels (1981). "On the Thermodynamic Limit of Photovoltaic Energy Conversion". Appl. Phys. 25 (2): 119–125. Bibcode:1981ApPhy..25..119D. doi:10.1007/BF00901283. S2CID 119693148.
Not sure how to calculate that...
If you were designing a set of LED's purely for plant growth, and making them in sufficient quantities that you can design your own dopant concentrations etc, then you probably wouldn't end up with full spectrum LED's.
[0]https://www.samsung.com/led/lighting/applications/horticultu...
The first step in the photosynthesis is the light absorption. The loss there is due to the limited light absorption (range 400 - 700nm). So maybe this step could be improved by using LED lighting dialled in to the range of the plant. The rest of the loss chain would remain mostly the same.
Doing the math with the wikipedia numbers
Plant growing in open field
===========================
0.53*0.7*0.76*0.32*0.6 = 0.05413632
==> 5.4%
Plant growing with LED lighting
===============================
Considering 100 % LED light absorption by the plant (which is an exaggerated hypothesis) and the 12% LED lighting efficiency
0.12*1.0*0.7*0.76*0.32*0.6 = 0.01225728
==> 1.2% efficiency
[1] https://en.wikipedia.org/wiki/Photosynthetic_efficiencySo take your 12% delivered Solar PV -> LED light to the plant ... and then apply 5% photosynthetic efficiency to that.
The efficiency of naturally-delivered sunlight to plants is of course 100%.
(I believe that's what you're actually calculating here, though your "As a comparison..." statement confuses the message.)
But, from the article, it seems like the biggest bill to be cut is the 7-figure executive salaries. They are obviously not worth it.