If 100km^2 is read as an area (not a square 100 km on a side), that's 1e8 m^2. Assuming an average of 500W/m^2 daytime insolation over 8 hours and an average conversion efficiency of 20%, that's about 1e9Watts (1 GW) average output over 8 hours, or 8GW-hr/day of energy.
Assuming 365 days/year, I get about 2.9e4GW-hr/year, or about 29 TW-hr/year.
Wikipedia [1] gives the total energy used by civilization in 2013 as about 157 PW-hr, or 157,000 TW-hr. The same source estimates electricity generation alone was estimated at about 19,500 TW-hr in 2013.
So 100km^2 of solar panels looks like it could produce about 0.15% of humanity's 2013 electrical energy production or about 0.0186% of humanity's total 2013 energy generation.
If the initial solar array dimensions should have been read as 100km x 100km, then its output would be 2 orders of magnitude larger. This would bring its output to 15% and 1.86% of 2013 electrical and total energy generation, respectively.
YMMV - these are only napkin calcs, not reviewed for error or fatal simplification.
Also you're right it's 100x100(miles even)² (https://www.inverse.com/article/34239-how-many-solar-panels-...)
US insolation is around 5.5 kWh/m^2/day. For a non-tracking array in Arizona it's about 6.5-7 [1]. Total energy collection would be almost 10^20 joules with 100% efficiency. US electricity consumption is about 1.4 * 10^19 J [2] so it would only require solar panels with 15.5% efficiency. Average is around 17%.
[1]: https://www.nrel.gov/gis/solar.html
[2]: https://en.wikipedia.org/wiki/Orders_of_magnitude_(energy)
If it's national, the 36x isn't that big of a factor. There are other locations with high solar input (africa, australia, middle east).
to challenge musk/solarcity claims about the 100km² solar farm being equivalent to the world energy output. Energy transport aside of course, just as a napkin calculation thing.