Where are you getting those prices? NASA was paying $3,60/kg for LH in 1980, so that's probably, what, $7/kg for LH today? Remember, this is LH, not gaseous - you not only have to cool it to extreme temperatures, but you also have to catalyze the conversion of orthohydrogen to parahydrogen - which is exothermic, yielding enough heat to nearly boil off everything you just cooled. NASA was paying $0,08/kg for LOX in 1980, so probably around $0,15 today. The Shuttle ET holds 630 tonnes of LOX and 106 tonnes of LH, so $836k.
The SRBs are 70% ammonium perchlorate, which is about $3/kg. 16% aluminum (about $1,50/kg), 12% PBAN binder (about $1/kg), 2% epoxy (about $5/kg), and an irrelevant amount of iron oxide. The total propellant was about 500 tonnes. Total propellant cost, $1,3m.
So the total propellant cost between the two, about $2m. To lift 27,6 tonnes of cargo to LEO, or $72 per kilogram. Now, people shouldn't fall for the fallacy that you just multiply that by how much a person weighs or a little more and that's the per-person cost to go to space - you actually have to launch many times more than a person's weight to get them there and keep them alive. But yes, propellant costs are not the key issue - if costs were close to propellant costs, rocketry would only cost about $25-100k to bring people to orbit in bulk.
Unfortunately, that's not the case.
Mind you, it's even possible to get significantly lower than that, but you can't rely on the rocket equation. And even if Space Elevator unobtanium existed, it wouldn't actually get you down to the levels one wants - there's no practical way to pump the climbing power up the tether, and beaming efficiencies with such a small receiver are unfortunately very low over such long distances. Much more practical is something like a Lofstrom loop - one might get power transfer efficiencies upwards of 50% or so. In such a case, you need about 70MJ per kilogram (19,4kWh). At industrial power rates of, say, $0,08/kW, that's a cost of a mere $1,56/kg. Sending people up in bulk might cost on the order of $800-ish per person in energy costs.
In both cases, though, it's not the propellant/energy costs that are killer, it's the hardware.You're asking structures to perform some borderline magical tasks in terms of the challenges that are put on them.
Anyway, enough Slashdot for now... back to working on simulating my caseless rocket design in OpenFoam and optimizing propellant combinations in CEA. ;)