It's no secret that the SpaceX Dragon capsule has a very impressive heat shield believed to be capable of direct lunar return. Official statements from SpaceX that they intend to add deployable landing gear and leverage the thrusters in order to land on land in the future prompts an obvious suggestion: if it can land on Earth, could it land on the Moon too?
The Dragon capsule has thus far been launched on the Falcon 9 booster, and although that booster is able to put 2473kg into lunar transfer orbit, after using the Draco thusters on the Dragon to enter low lunar orbit the total mass would be under 1876kg.. this seems a bit light for a crewed configuration, especially when you consider that only 1422kg of it could be returned to Earth. And that's just lunar orbit.
We need a bigger rocket, and the official SpaceX plan right now is called the Falcon 9 Heavy. One should not be confused by the name, the F9H is not "heavy lift" in the sense often used by space advocates and policy makers - who should more correctly be using the term super heavy lift. So how heavy is the F9H? Comparison is usually done in terms of lift to LEO, but for our purposes lift to LTO is more interesting at 10622kg. A slight improvement on the Delta IV Heavy at 9984kg.
Now we can imagine a Dragon-lander flying direct from lunar transfer orbit to the surface, it would have a mass of 3766kg when it landed which is quite respectable. For a cargo flight this is fine, delivering 2737kg of payload, but it's unlikely the vehicle would have enough fuel left to attempt an ascent.
Having determined that a single stage direct descent vehicle is unlikely, we're now forced to choose a mission mode. The size of the launch vehicle has already dictated that LEO should be bypassed, so our choice comes down to lunar-orbit rendezvous (the mode used by Apollo) or lunar-surface rendezvous, aka, refueling on the surface. So much has been said about LOR already, so let's run the numbers for LSR.
Having landed a crewed Dragon-lander on the surface, and assuming no fuel is left, we would require 6014kg of propellant to return to Earth. This is not too bad, at 3 fuel landings, but we can do better. If we can carry just 540kg of fuel in reserve we can eliminate the third fuel landing. Another alternative is to throw 338kg of payload out.
Of the 2737kg payload delivered, we have to determine how much is needed for the crew and their supplies, and how much can be fuel. The pressurized volume of the Dragon is 10m^3 requiring 11.839kg of air to fill. Without an airlock we may wish to cycle that a few times, so let's say 118kg total. Next we need a one week supply of oxygen candles at 25.83kg per person, and LiOH to scrub the CO2 at 52.71kg per person. Finally there's food and water at 45kg per person. For a total of 488kg for a crew of three. Too easy! This leaves 1708kg for spacesuits and equipment.
Once on the surface, the crew would vent the chamber, get out and refill the fuel tanks. Having gravity, transferring the propellant is well understood. Return to Earth would be direct, with no need to enter lunar orbit or perform a rendezvous. As no parts fall off the Dragon-lander on the way it could be fully reusable, providing a stepping stone to in-situ produced propellants.
I estimate a Falcon 9 Heavy / Dragon-lander cargo configuration would cost around $40k/kg to the lunar surface. With the two fuel emplacement flights, this makes crew transport something like $130M/seat for crews of 3, but you could conceivably get that down to $55M/seat if you were delivering 7 at a time - most likely to some kind of base as they would have reduced volume for equipment.
Most of my calculations were done with this rocket equation calculator and I used an inert mass fraction of 0.15 for the lander.