Why "a" huge complex when "several" not-so-huge complexes would be just as good?
Because "not so huge" in this context basically means something at least as large as the ISS. It just works out this way mathematically: solar energy density near Earth is about 1360 watts per square meter. A solar panel that gets around 50% efficiency in conversion (e.g. fictional, futuristic super-panels that don't exist yet) would therefore net about 700 watts per cubic meter. For an orbital solar platform to produce commercially viable power it would need to have an output of around 2GW (even ignoring losses due to transmission from orbit to ground).
This means your solar power station -- or stations, in that case -- must support solar arrays with a surface area of 2,857,142 square meters. By comparison, the International Space Station's solar arrays have a surface area of around 3,000 square meters. By further comparison, the total surface area of the Sears Tower is about 420,000 square meters; if you had one Sears Tower's worth of solar arrays in geostationary orbit with super-fancy futuristic high efficiency solar cells, it would have a power output of about 294MW. Because of the huge costs involved, that's not even CLOSE to being competitive with conventional power sources; you'll need at least 8 more of these huge structures if you want to break into the market. But let's say 300MW is a "starter" power plant. Where does this get us?
It's been a LONG time since I looked at the actual numbers, but I once calculated a figure of about 1kg per square meter of solar cell, excluding power conditioning hardware. Which means your 300MW orbital solar array has a mass -- FOR THE ARRAYS ALONE -- of 420 tons. Nobody knows the exact cost of a typical HIIB launch, but let's be generous and assume they're secretly competitive with SpaceX (only in their wildest dreams, but let's pretend that for now) and give them a cost to orbit of around $5,000/kg.
So the cost
just to launch the solar panels would be $2.1 billion. And that's before you factor in the cost of the power conditioning hardware, the microwave transmitters, the control systems, the various components of the truss structure, support structure, stationkeeping hardware, etc. And all of THAT before you factor in the cost of the receiving station, which is basically a whole second power plant of its own. This is, in the end, about $10 billion for a power plant that produces only slightly more electricity than a coal-fired plant that could be setup for $400 million. And the really shitty part is, you could actually eliminate most of that cost by simply NOT launching the solar panels into orbit; most GROUND-based solar power plant in the 300MWe range cost between $500 and $700 million to build. And remember that this is based on the assumption that Japan can achieve launch costs that even SpaceX can barely sustain; the real figure, once you include development costs, and the actual cost of Japan's launch vehicles, would be in the neighborhood of $50 to $100 billion.
So orbital solar energy as a commodity IN AND OF ITSELF makes no economic sense. It does exactly what ground-based solar does, it just does it way less efficiently and way more expensively. It is, on the other hand, a relatively cheap way for a very large facility already in place to take advantage of its surplus natural resources -- e.g. an abundance of intense sunlight -- and export that resource back to Earth. That is, people who ALREADY LIVE IN SPACE would find orbital solar power to be one of their most convenient export commodities.
Basically: orbital solar power will never be profitable to anyone who doesn't already live in orbit. It costs too much to ship all that hardware up the gravity well -- and is too risky to leave it all unattended for years at a time -- for it to be even slightly practical from our end.
