A slightly revised comment. Bolded words have been added to reduce confusion...
It occurs to me that some people might be sceptical of my statement that solar power satellites (SPSs) could satisfy Earth's total energy needs many times over. Here's a quickie calculation that proves the possibility -- in principle. In practice of course, we would scale it way down, at least at first.
Imagine a band of solar panels completely encircling the Earth in geosynchronous orbit. The band is 10 km wide. This is a mind-bogglingly enormous structure of course. But we're just doing a back-of-the-envelope design, so bear with me.
The surface area of this band is easy to compute:
A = 2 * pi * r * w
where A is the surface area, pi is the mathematical constant, r is the radius, and w is the width of the band. Geosynchronous altitude is roughly 42,000 km from the center of the earth, so r = 42,000,000 m. Our imaginary band is 10 km wide, so w = 10,000 m.
Hence A = 2 * pi * 42e6 m * 10e3 m = 2.6e12 m^2.
Solar irradiance on Earth is roughly 1400 W/m^2. But our cheap solar panels (manufactured on the Moon) might have only 15% efficiency. Plus not all parts of the band will be facing the Sun squarely at the same time; however, both sides of each panel would have the ability to catch whatever sunlight is shining on it. As a rough guess, we need to cut the average efficency in half again. So the power would be
P = (2.6e12 m^2) * (1400 W/m^2) * 0.15 * (1/2) = 277 terawatts.
Currently Earth is consuming 18 terawatts; this includes every source of energy like coal, oil, etc. So our orbiting solar power band would be producing 15 times more power than all of Earth is presently using.
We won't need this much power, not at first. So let's scale it down: let's cut the continuous band into 10 km x 10 km squares, and space them 60 km apart along the circumference of GEO. So only 1/7 of GEO's circumference would be occupied by solar panels. This scaled-down system would produce merely twice as much power as all of Earth is using now.
How many satellites would be necessary? This is easy too: the circumference of GEO divided by 10 km + 60 km = 70 km, or
2 * pi * (42,000 km) / (70 km) = 3800 satellites
This seems feasible.
A single orbiting 10 km x 10 km solar farm seems massive; wouldn't it block our view of the stars? No, not at all. At geosynchronous altitide (35,000 km), a farm 10 km wide would subtend
arctan(10/35000) = 0.016 degrees
As seen from Earth, the full Moon is half a degree wide. From the surface of the Earth, each solar farm would seem less than 1/60 as wide as the full Moon; if you looked up at night, these farms would form a dotted line of tiny shining pinpoints from horizon to horizon. The glare could upset some astronomers, but by the time we can build the SPSs, my hope is that the forefront of astronomy will have moved to giant telescopes in space, beyond the Moon (or on it).