That sounds like an interesting project.
It seems to me you could approximate the power requirements based on the typical parameters of more conventional designs, since the basic CO2 laser physics would still apply.
In other words, say your active length would be 33 cm; a conventional dc or rf driven design might give you 20 to 30 watts out at 10 to 15 percent efficiency...so, picking something in the middle...maybe 200 watts in to get 20 to 30 watts out.
I wonder what the impedance would look like?
Before it ionizes, you'd have a high impedance, and consequently a high E field on the cavity axis, which, it seems to me would facilitate ionization; after which the impedance would be much lower.