The problem is that the current design for the lunar Starship doesn't have enough propellant left over to leave lunar orbit, let alone enter Earth orbit (and it has been stripped of its heatshield, so no aerocapture). Going back on a Starship requires first delivering fuel for that Starship to lunar orbit, which adds another ~8 launches to each mission.
They could even send the same kind of vehicle ahead of schedule to the moon as a backup for when the astronauts are going home.
And just saying 'its complex' therefore lets spend 40 billion on development and then limit our options by paying 3 billion per launch is an insane plan.
Overall complexity is reduced by not having those two legacy systems.
Basically you use Falcon9 to get in and out of LEO using Dragon.
A Falcon Heavy can push a larger space craft into LEO and it saves weight by the fact that a LEO-to-moon vehicle doesn't need heath shield and other stuff which makes a Dragon capsule heavier.
And with Falcons being cheap you could even send a whole bunch of LEO-to-moon vehicles straight to the moon surface, as well as just dump fuel tanks there.
SpaceX has a lot of these and can build them quite cheaply, so the mission will have a lot more built in safety. E.g. if you got an extra lunar ascent vehicle sitting on the landing site already before astronauts arrive, you have extra redundancy.
If you actually did a clean sheet architecture, there is literally zero reason to do SLS/Orion.
Starship is one on the critical path anyway. The only difference is you go from the capsule to Starship in LEO instead.