Sending humans in a SpaceX Dragon v2 capsule to EML-1 or 2 is a worthwhile possible step in a 100% commercial return to the Moon. The SuperDraco thrusters to be included in the sidewalls of the crew Dragon capsule are more than capable of performing the trans-lunar injection burn, as well as station keeping at the Lagrange point, rendezvous with any preemplaced assets - such as a lander - and returning the crew to Earth.
The total delta-v for such a bare bones mission to EML-2 is a mere 4835 m/s. EML-1 is similar. This is a "quick transit" 4-day trip up to EML-2, so the crew spends less time in the radiation belts. A good estimate of the dry mass of the Dragon v2 is 8000 kg. Using an isp of 320s, the initial mass in LEO is just 37344 kg, or 68% of the maximum payload mass of a Falcon Heavy.
Just going to a Lagrange point with a crew on Dragon would be a momentous achievement and could be done for a mere $150M. However, it is just the first step.
If we fill the remaining Falcon Heavy payload mass with fuel, the total delta-v available to the Dragon becomes 6050 m/s. This is sufficient to go from LEO to EML-2 to low lunar orbit and back to Earth, with significant margin for maneuvering and rendezvous, if required.
What might the crew in the Dragon rendezvous with at EML-1, EML-2 or low lunar orbit? Using the same SuperDraco thrusters to take the slower 9-day transit - a delta-v of just 3470 m/s - 18.2 tons of payload can be delivered to EML-2. This payload can loiter for months waiting for the crew to arrive. Up to 14.8 tons can be deployed from EML-2 to low lunar orbit when required. Up to 6.2 tons can be landed on the surface if the lander is taken via low lunar orbit, or up to 8.1 tons if taken directly from EML-1 or 2.
This is all possible because I have avoided two pitfalls of lunar return architectures that have become very common in recent years.
Firstly, I have ignored the possibility of using the second stage of the Falcon Heavy to perform any part of the trans-lunar injection burn. A simple trade study shows that it is not advantageous when you have sufficient thrust on the payload - which you must have to do later maneuvers - so I'm baffled as to why people keep considering it.
Secondly, I have not used any high isp propulsion such as LH2/LOX or CH4/LOX. Although this may become preferable in future lunar architectures (especially if propellant made on the Moon becomes available), it is currently an additional expense which does not provide significant advantage to justify its cost.