EDIT: Which has now been replaced by the patent application, for the general confusion of everyone...
[1] http://patft.uspto.gov/netacgi/nph-Parser?patentnumber=90858...
Edit: My mistake, I didn't realize that this wasn't the normal concept of tether out past geosync.
3% to 30% is a 10x savings, because it's a ratio.
E.g., if it takes $1,000 of fuel to get a unit mass into orbit, then a 3% ratio means $33K, and a 30% ratio means $3K.
I am always skeptical of "massive upfront investment but lower per-unit costs," especially for space-launch. It's the way most of us learn about space, because sci-fi novelists talking about space have to explain what finally got man into space, and a massive infrastructure seems such a nice logical explanation. But Musk has shown that Plain Old Boring Rockets can be done much more efficiently than traditionally done. And that doesn't make a good sci-fi backstory.
If we assume that SpaceX cuts the overall cost of launches to 1/10th mainly by reuse of the first stage, that is indeed a 10x savings, but not on fuel (in fact, SpaceX uses more fuel because they need to be able to slow the first stage).
0.3% of x is the same as 3% of x/10.
So my point was that even with SpaceX numbers, where fuel will (once they sort out reuse) make up substantially more than what it currently does (because they're aiming to massively slashed the non-fuel costs), a 30% cut in fuel costs will "best" case cut launch costs by about 1%.
If these guys are going to achieve substantial cost cuts it will need to be elsewhere unless Musk' has been massively misquoted or is totally wrong about the relative cost of fuel vs. the rest.
(I'm not going to bother being precise as all of these are handwavy estimates from people anyway).
E.g. consider that Musk a few years ago estimated the fuel and oxidizer costs for a Falcon 9 v1 at ca $200k per launch, with a list price for the launches at $54m to $59.5m. Cutting the cost of the rockets themselves on the other hand, whether by reuse or making them smaller/cheaper will matter much more.
I guess whether eliminating the first stage entirely vs. reuse will matter much will depend on how many times recovered stages can be reused and how much it will cost to prepare them for reuse, as the potential fuel savings are basically rounding errors.
I agree 100% that the economic viability of a "space elevator" like this is basically non-existent, even though it's a really fun thing to think about.
Also, is 30% of fuel costs really going to be worth it? According to [1] Elon Musk claimed fuel is only 0.3% of the cost of current rockets. Even if we assume SpaceX slashes the overall cost to 1/10th without touching fuel costs, that still only brings fuel up to 3%. 1% additional savings doesn't seem like it's going to do much to help finance a 20km tower.
Of course if single stage to orbit planes are viable from it, then that may make it more attractive, but it seems peculiar to focus on the fuel cost.
[1] http://www.space.com/21386-spacex-reusable-rockets-cost.html