http://keith.seas.harvard.edu/papers/148.Holmes.Keith.Contac... is probably the most advanced current geoengineering proposal currently; the estimated cost is US$60 per tonne of CO₂ removed, proposing to spend C$140M to construct a facility that removes 1 megatonne of CO₂ per year. (One of the authors constructed a small-scale prototype a few years earlier.) If this is correct, a facility to remove all of the 29 billion tonnes of CO₂ emitted by human activities annually would cost C$4 trillion to build (about one month of global GDP) and cost US$1.7 trillion per year to operate (about 5% of global GDP, comparable to the total amount currently being spent on energy.)

That is, unless Keith's analysis is wrong, low-risk geoengineering to completely remediate current CO₂ emissions is clearly economically feasible, but not as a skunkworks project — it's a project of a size comparable to the entire existing energy industry, so it would need to be supported by the full influence of the current Establishment. It clearly cannot be done as a market-driven project.

These costs, already somewhat pessimistic, should come down a bit as energy becomes cheaper with the proliferation of photovoltaic farms. Indeed, much of the operating cost is the energy used, which would have to come from non-carbon-burning sources.

Many of the other geoengineering approaches mentioned — cloud doping, ocean fertilization, stratospheric acid, space mirrors — would probably be cheaper. Unless they crash the Earth's climate system!