It seems to me that any plan to deploy solar power satellites (SPS) on a scale to largely replace fossil fuels must seriously consider Gerard O’Neill’s strategy of building them largely from lunar resources instead of hauling all of that mass up from the bottom of Earth’s gravity well. I’d expect the first stages of SPS deployment to follow something like the roadmap in John Mankins’s 2014 The Case for Space Solar Power, which envisions a five stage program of design reference missions building successively larger satellites from eight kinds of identical modules, none massing more than 600 kg, starting with small demonstrators and building up to a pilot plant with a cost he estimates at US$ 5 billion, with a total budget to complete the first full-scale powersat at $20 billion, which is peanuts compared to what NASA is dumping into the ocean with SLS at US$ 4 billion per launch according to the NASA Inspector General. These figures assume all components built on and launched from Earth, with the full scale SPS requiring 500 to 1000 launches of payloads between 10 and 20 tonnes and a reusable launcher with cost to low Earth orbit between US$ 300 and 500 per kilogram. SpaceX is estimating figures for Starship much lower than this, but “we’ll see”.
But rather than hauling all of that mass (and the propellants needed to lift it) all the way from Earth, isn’t is more practical, when there’s, from an elemental composition standpoint, this huge ball of oxygen, silicon, iron, and aluminium just sitting there less than 4 km/sec delta-v from geostationary orbit (that’s lunar surface to GEO; low lunar orbit to GEO is just 2 km/sec). If, as O’Neill suggested, you launch regolith from the Moon’s surface with an electromagnetic mass driver and refine and fabricate in lunar orbit, propellant requirements to transport components to GEO are a fraction of bringing them from Earth (2 vs. ~13.2 km/sec). And, when O’Neill was proposing this, he didn’t know of water resources on the Moon which can be processed into hydrogen/oxygen propellant.
Now, building out the infrastructure to do this sounds fantastically ambitious in an age when it seems that in projects that actually involve atoms as opposed to bits, it takes almost forever to accomplish almost nothing, but so is launching and recovering 315 Starship-class vehicles per day, building the infrastructure and propellant production this would require, and building new Starships at a rate of one every three days or faster.
It would be interesting to work out the mass budget of lunar resource collecting, processing, and manufacturing required to build operational SPS at a given rate, then calculate how many Starship launches, including refuelling tanker flights, it would take to deliver that to the lunar surface and orbit, and then the rate of support launches that would be required after the infrastructure was in place (for repair and replacement parts, crews and supplies, and high-cost, low-mass components and materials [for example, microchips and rare earth metals]).