Elon Musk usually seems to do the math instead of just trusting his gut. That’s why I find it quite interesting that he thinks that a colony on Mars makes more economic sense than a space habitat.

There are a couple of things that I think are pretty bogus. One is space mining, another is space solar power. I mean if you calculate how much it costs to bring either the photons from space solar power back to earth or the raw material back to earth, the economics don't make sense.

—Musk at a Stanford Entrepreneurial Thought Leaders Lecture

I’m wondering what kind of assumptions he made. If you think of building the solar panels and mining equipment on a planetary surface, I absolutely agree, it’s not viable since it’s just not realistic to bring down launch costs that far. However, if you’d manage to establish a whole industry in space (i.e. orbit around Earth or Sun), the economics might change. If it became much cheaper to mine resources and build stuff in space, cheaper than doing that on a planetary surface, you wouldn’t have to bring back much raw material, only finished products. If it became cheaper to build ten solar panels in orbit than to build one on the surface, the terrible efficiency of transmitting the energy down from orbit wouldn’t matter anymore.

Meanwhile, Elon Musk is focussed on Mars:

If we can establish a Mars colony, we can almost certainly colonise the whole Solar System, because we'll have created a strong economic forcing function for the improvement of space travel. We'll go to the moons of Jupiter, at least some of the outer ones for sure, and probably Titan on Saturn, and the asteroids. Once we have that forcing function, and an Earth-to-Mars economy, we'll cover the whole Solar System.

—Musk in an Aeon interview

I’m kind of curious why he thinks that an Earth-to-Mars economy will be more economical than an Earth-to-Space economy. Since you’d want the off-earth economy to not only import, but also export goods, Musk seems to assume that building a productive colony on Mars plus the launching costs from Mars are actually cheaper than building a productive colony in space. (Exports from Earth shouldn’t be that different since you can aerobrake a lot in the Martian atmosphere.)

In addition to some basic considerations, is it really that much harder to reinvent everything for a zero-g environment than for a Martian environment? Maybe it is. And maybe it isn’t.