I'm sick of anti-Elon hysteria, and I while I respect differences of opinion, I downvote FUD that's based on misinformation. SpaceX commercially bid for, and won, a launch services contract, which they are delivering at a considerably lower cost than the competition. Being paid for a commercial service is not the same as being subsidized -- And in any case, NASA is not their major source of revenue.

But if your question is sincere: the cost of the fuselage (and engines, and communications and control systems, and the considerable amount of tech that goes into rockets) are considerable. Scattered around various tweets, Elon has indicated that the breakdown of costs for a (pre-Block 5) Falcon 9 are as follows:

- Payload fairing: $6M

- 2nd stage: $12M

- 1st stage: $24M

- Everything else -- propellant, operations, permits, etc.: $6M

So recovery of the 1st stage would save about 50% of the launch costs, if such a recovery were itself cost-free. Obviously it does require some extra hardware, propellant, and operational complexity -- but as long adds less than $24M per launch, it's a net positive. I've seen no suggestion that it adds more than a few million per launch.

If you want to increase the number of rocket launches, you need to increase the number of rockets. We will always launch them as fast as we can build them, and it’s probably cheaper to develop reusable rocket tech than it is to build a single additional rocket production facility. It makes every production facility afterwards that much more efficient.

It was a great economic move if your goal is to increase the number of launches as quickly as possible and make spaceflighy routine.

Liquid engines are far more flexible that solid, and due to the Falcon 9 design using nine engine, SpaceX has built over 500 Merlin engines. That gave them an assembly line approach instead of a hand building approach, which led to both massive cost reductions, and continuous improvement processes.

That nine engine design was also key to mastering retro-propulsion, because it allows for the deep throttling required.

And if you don’t understand the quantum leap reusability is, you don’t have any understanding of launch economics. Solid rockets will never be economic because they aren’t reusable. Every launch, SLS is going destroy $500M of hand built custom rocket parts (ignoring another couple billion in development and operating costs).

If the BFR cost $500M each, and can fly 100 tines with only minor refurbishment, that’s only $5M per launch. With refurb, operating and fuel costs, total launch cost would be near $10M, for a cargo capacity 59% bigger than the SLS.

Reusability brings flight costs close to fuel costs, which are trivial. The Saturn V cargo cost was about $10;000 per pound, the Shuttle $40,000 per pound, commercial rockets were around $5,000-$10,000 per pound. The Falcon 9 is under $1,500 per pound, the Falcon Heavy under $1,000 per pound, and if the BFR concept works it will cost between $20-$50 per pound.