If we could build a robust fully-reusable space plane that can go Mach 15 free from the nastiness of the rocket equation then we already practically achieved inexpensive space access :). If you look at the rocket equation - it is the first 15 Mach that require to burn almost all this mass of fuel through all these big engines pumped from all these big tanks. Once you're going 15 Mach - to get to the 30 Mach for the payload of N kg, you'd need to burn only ~N kg of LH2+LOX (4.5km/s nozzle speed) - so not much gain would come by replacing rocket propulsion with anything else on the second half of the acceleration. The first half is the problem.

By not requiring full orbital velocity, we are freeing ourselves from the particular nastiness of the rocket equation on the surface of our planet. Our gravity well is just deep enough, that SSTO is just beyond the edge of what we can achieve with chemical rockets. I proposed Mach 15 because it's significantly less than the Mach 20 result, but in this case it's clear that I still chose a number too high.

An air breathing craft that can get to Mach 10+ would easily drop the cost of getting to space to 1/10th what it is today and let do things like have some reserve fuel in case you messed up the landing and want to come around for another pass. If you can hit Mach 20 then LEO starts to take less energy than flying around the world.

there are 2 issues with oxygen from the air:

1. the air is moving fast relatively to the aircraft frame of reference. To burn it, the air should be slowed down in the aircraft frame of reference. That means acceleration of that air in the Earth frame of reference. It isn't a noticeable energy loss below Mach 2, yet above the Mach 2-3 it starts to impact the efficiency of the air breathing scheme to the point that rocket engine carrying oxygen with it (ie. accelerating it in the Earth frame of reference and keeping it non-moving in the aircraft frame of reference) doesn't look that less efficient, and the higher the speed the less the efficiency gap. Scramjets (no slowing down of the air) while seemingly fixing that problem face the other side of the same issue - trying to impact even small amount of additional momentum on the already fast moving air (in the aircraft frame of reference) requires an unproportionally increasing (the same v square) amount of power.

2. due to composition of the Earth atmosphere, to burn 1kg of oxygen, the aircraft need to pump through engine - ie. accelerate as described in the point 1. above to the speed close to its own in the Earth's frame of reference - 5 kg of air.

The jet propulsion itself doesn't care that mass m being thrown out of the nozzle with velocity v consists of - H, O2, H2O, steel balls or foam - doesn't make difference as long as it is of mass "m", and the aircraft needs to come up with that mass somehow - carry it with itself or gather from outside (and accelerate the mass or try to impact momentum on the already fast moving mass).

Take the points above, add the simplicity and low weight of the rocket engine in comparison with air breathing engines and you will see why rocket engines are dominating the arena.

2. Scramjet's have already been demonstrated at over Mach 9.5 and are still more efficient than rockets at those speeds. (aka more thrust per unit fuel.)

3. Scramjet's are actually fairly simple designs compared to a traditional get engines. (Modeling and testing them however is much harder.)

Granted, there are plenty of downsides which is why we have not built anything like that. But, the issues are far more in line with heat dissipation and drag vs. limits on the basic physics. Still the main limitation seems to be the far lower thrust-to-weight ratio.

that exactly the basic physics limit of the scramjet that i was talking about : "the other side of the same issue - trying to impact even small amount of additional momentum on the already fast moving air"

http://en.wikipedia.org/wiki/Space_elevator_economics

A better example might be fancy chemicals. Rather than purchasing a "very expensive machine" to ensure a specific environment for reaction, it might be cheaper to send the materials to zero g for processing. I'm not a chemist, but watching astronauts play with bubbles and flames in space makes me think you could get a lot of precision by modeling everything as perfect spheres. So, the reactions are simper to model, easier to get accurate and precise outcomes.

Delivery of all goods is simply a drop shipment.

What I'm proposing isn't made of unobtanium. Spectra has the required tensile strength.