I was reading this question and realized that I didn't actually know enough to give the answer that I wanted.
Obviously in the near term, all our raw materials come from the Earth. The only other place we've been is the Moon, and we haven't been there lately. So if we built a space station with gravitational rotation, we'd have to build it with Earth materials at the moment. The thing is though that in the long term, I don't believe it makes sense to lift materials from Earth.
Lifting materials out of the Earth's gravity well is energy expensive. It seems like it would be much easier to get materials that are already in space and redirect them to where we want. For example, we could collect materials from the asteroid belt and direct them to the Earth/Sun Lagrange-5 point. Starting to do this would be expensive, as it requires flying out to the asteroid belt. But it seems like it should be easier to maintain once started. Also, this would save mining and other costs on Earth.
I would think that we'd mainly do this with self-replicating miner robots controlled (loosely) from Earth. I'm assuming that we'd make similar robots for large scale space construction regardless, so the only difference is in energy costs. Energy is plentiful in space, as you can build arbitrarily large solar panels. So I'm figuring that the asteroid project will be self-sustaining once started.
The question remains: how much stuff could we get into Earth orbit for the start-up cost of initiating the asteroid project? Ideally the answer should be referenced in terms of population capacity. For example, "we could either build space stations for 10,000 people or build an asteroid harvester" would be an example of the kind of answer that I'd like to see.
Feel free to neglect the cost of providing atmosphere, water, and food. I'm going to assume that all those things come from the same place in either scenario. This will probably be the Earth initially but may be someplace more exotic later (Venus, Saturn's rings, a Jovian moon, etc.).