[bediger4000]

Bottom line, with some solid but not necessarily heroic engineering, the cost of an orbital data center could be as low as three times that of the comparable terrestrial one.

"Could be as low" means the 3x cost is based on optimistic assumptions. When does something 3x as expensive ever get chosen? Usually we're concerned with shaving single digit percentages of costs.

[__patchbit__]

As a rule of thumb use the mass to orbit launch cost curve and cadence of SpaceX vehicles to project forward the orbital station solar powered AI datacenter cost curve. It will be highly unlikely to cost $42 billion and not even one AI datacenter ever reaches orbit.