Why not solar energy on earth ??
Space is hard to get to. Why not just power server farms from solar cells on earth?
Simply put, massive habitat destruction.
150,000 Terawatts of solar energy hits the Earth. Some is reflected, some is absorbed by clouds. The energy actually reaching the surface of the earth is called insolation, and is usually measured in kilowatt hours per square meter per day. A fixed plate solar cell in New Mexico gathers 7 kwhr/m2day, or an average of 290 watts per square meter, tilted towards the sun. Because of the 32+22 degree winter tilt, it actually shadows 1.7m2, for an average of 170 w/m2 of occluded ground area. This is best case in winter (when energy demand peaks), and assumes that there is some way to store power over long periods - if New Mexico is covered by clouds or snow, the output drops to almost nothing.
Assume a 6% efficient energy conversion from sunlight to the customer. That includes manufacturing, maintenance, energy storage, conversion, and grid transmission. This is also optimistic.
For each square meter of New Mexico real estate, we can deliver 10 watts to the customer. This is a far cry from the 1366 watts per square meter in space. On that basis alone, it is cheaper to orbit the data centers.
But wait, there's more. Oil may or may not be running out, but with the United States burdened by a huge international debt, and growing economies like China demanding a larger share of world production, we may soon be unable to import foreign oil, and forced to develop an electric transportation infrastructure. The total energy use of the United States and Canada averages 4 terawatts, mostly for transportation. 4 TW divided by 10 W/m2 is 400,000 square kilometers, or 155,000 square miles, bigger than New Mexico, about as big as California. Can you imagine how much destroyed habitat that represents? Paving an area the size of California with solar cells would be an ecological nightmare.
It would also be an economic nightmare. A solar cell panel designed to survive the outdoors for, say, 50 years is a far more complicated object than a patch of paved road. The entire interstate highway system is 75,000 kilometers long. Assuming an average paved width of 100 feet, 30 meters, the area of the system is 2250 square kilometers, and that cost 130 billion dollars, or about 60 million dollars per square kilometer. If the solar cells, converters, storage, and transmission systems cost only as much per square kilometer as the interstate highway system, the whole thing would cost 23 trillion dollars. That is years of economic output, meaning years of privation and disruption. No thanks.
Imagine how much it would cost to keep the interstate highway system optically transparent ...
Alternative: Server Sky
If some tiny fraction of that enormous cost was spent on research to get server-sat thickness down to 5 microns, reducing the weight to 300 milligrams, and reducing launch costs to 1000 dollars a kilogram, then the cost of orbiting a 5 watt, 85% availability server-sat would be 30 cents. Orbiting 4 trillion watts of server-sats would cost 280 billion dollars.
Development would not stop there, of course. A system such as the launch loop could loft server-sats for much less. Launch costs would essentially be free, relative to manufacturing cost. If server-sats were made of lunar silicon, and used for space solar power, they would be far heavier (perhaps 500 microns thick with coverglass, and 30 grams), but "closer" to deployment in the LaGrange positions, from electrically-powered lunar launchers. The environmental cost would asymptote towards zero, and the system would provide its own launch energy.
The only thing holding us back from an abundant energy future is faint-heartedness. This will get done, the only question is whether Americans have the courage to participate.