In the fullness of time, we will move terawatts from space to earth, selling industrial "raw material" harvested from space. Transporting energy is expensive and dangerous, and the customers (utility companies) are unimaginative and conservative as hell.
Let's be a little more clever about this.
In the late 1700's, the farmers of western Pennsylvania and the Ohio valley were separated by weeks of arduous travel from the millions of customers on the eastern seaboard of the new United States. No railroads, no canals, roads that were little more than trails for pack animals.
Transporting 600 kilograms of rye to eastern markets required four pack mules. That rye sold for $6, and did not pay for the journey.
Instead, the clever farmers fermented that rye into 30 liters of whiskey, which they could move on one pack mule and earn a $16 profit after expenses. The left-over mash was slightly alcoholic and depleted of carbohydrates, but it still had lots of protein and made great animal feed. For chickens it was better than the original rye. Drunken chickens are happy chickens and make bigger eggs with the concentrated protein.
So why transport electricity when we can transport products worth far more and costing far less to move?
Google spends $300M a year on electricity, which it transforms into information products, search-related advertising selling for $30,000M a year. That's lucrative, but we can do better.
India produces world class information technology services, using even less electricity to produce revenue. India has half a billion unconnected people in the countryside, many of whom could also learn to do programming and support work. What if we could use space power to connect a trillion dollars a year of IT services to the world?
The silicon chips doing the actual work in google's data centers weigh less than 10 grams per kilowatt. The packaging, from circuit boards to buildings, weighs tonnes per kilowatt. In vacuum and microgravity, we can dispense with almost all the packaging. We can launch silicon cheaper than we can wrap a building around it.
If we compute in space, and send results with millimeter wave radio, we can move gigabits per second with a few watts of radio energy. Someday, when we learn to steer and phase-lock vertical-cavity-emitting solid-state lasers without lenses (Akhil? We need your BRAIN working on this!) we can move even more data with less power and weight.
That's the direction server sky is taking. I realize I am talking heresy in the Holy Church of Space Exploration, but I've been waiting for cheap launch and space solar power for 40 years, and the Space Messiah seems to be busy elsewhere.
Server sky won't happen right away either, but it can be built with existing technology and launched with one Indian PSLV launcher. Graduate students at Rajiv Gandhi University could design it, and a company like Tata or Reliance could pay for it. It would need chips from Intel, but about 10% of Intel's engineers come from India. I bet they could convince U.S. management to support this.
It's time to stop beating our heads against a wall that doesn't move. There are products we can sell now that can earn big profits and grow rapidly. If an old fart like me can think up server sky, what can all YOU smart people figure out?