Yes, the problem is likely not in getting enough carbon (though a kiloton of steel would require several tons of it), but rather having a mass production process advanced enough to precisely control it. Almost all heavy equipment would have to be bootstrapped on Mars, mostly from the inevitably subpar local materials, and ith access to energy that's worse than on Earth: no fossil fuels, no hydro-energy, much less sunlight.
Everyone who's seriously considered space colonization has come to the same conclusion that Eric Drexler did -- you need to have some kind of system that can make absolutely everything with as small a population supporting it as possible.
You've got the problem that there's nothing that could manufactured on Mars that would be worth bringing back to Earth. If a Martian colony was dependent on Earth for anything it would expect to get its resources cut off at any time, and even if you can get spare parts and stuff from Earth the turn-around time counting the synodic period and transit time will always be several years. See
https://en.wikipedia.org/wiki/The_Martian_Way
I think it could be possible with some combination of synthetic biology, fermentation, flow chemistry, 3-d printing and such. It's a good northstar for research into "advanced manufacturing" which could come in handy here on Earth.
Honestly I think one of the possible premier uses of orbital (though not Martian) resources would actually be agriculture. Limiting biological contamination and maintaining sterile environments, unlike other industries, can produce a value-added product compared to the inputs.
Also there's at least a plausible mass trade off - a space borne habitat structure doesn't need to support its own weight against gravity, so you might be able to trade favorably on the launch costs (e.g. grow crops in a big inflatable dome under hydroponic conditions). Certainly it would make enforcing quarantine easier.
Large structures and LEO do look like the closest you could come to the LEO dream. The atmosphere for a baby Bernal sphere that has a usable area of 35 acres would take about 15 starships to send up which would be like getting 1 starship load to the moon. Building something like that which is a simulation environment for Mars might be cheaper than going to Mars.