(1) The "small number of trains" was in the context of not wanting to extend to Las Vegas (which I don't think we can see supporting anywhere near that many trains).
(2) Well, let's presume that number is about right in terms of frequencies, and that those 39 departures would have about a 2:15 runtime (80 MPH average speed; getting up to 90 MPH average speed doesn't get you under 2:00, so this seems reasonable). 39 departures spread over 16 hours means that you're going to have a peak rate of 3x trains per hour in each direction, so you're probably looking at needing about 15-20 trainsets. To get 200 pax/train on average (which is about what 15,800/day comes to) you'd probably need closer to 400 seats per train...so you're probably looking at about a 7-8 car set, plus two locomotives, for the service. If each set is about $40m, this would come to $600-800m for equipment. This would leave about $1bn for the rest of the project...
(3) ...which would probably go to some mixture of the two stations (likely with about a half-dozen tracks given those frequency levels), an equipment yard capable of storing those sets, plus the cost of getting through the mountains heading into Vegas. I stand by the $10m/mile estimate for the desert run, but there will be additional costs when dealing with a mountain range.
I write this presuming, of course, that Palmdale-Victorville isn't included. If it is, that's probably another $400-600m for the tracks, plus a slew of additional trainsets (since runtime increases).