Here's why Elon Musk is more likely to be right: because fuel is cheap.
The cost of fuel and the cost of fuel tanks is an insignificant part of the cost of an orbital launch, around the 1% level. The major drivers of cost are overall system complexity and manufacturing cost of the engines. And here's the big problem for a Skylon spaceplane, rockets are fairly simple systems whereas hypersonic airbreathing engines are extraordinarily complex and difficult. And if you can manage reusability on your launcher then the ordinary rocket engine wins hands down.
The reason why the jet engine won out over the propeller in civil aviation is not because of the higher thrust or better performance of the jet, it's because of lower operational costs. A jet powered aircraft requires less maintenance per passenger-mile than a propeller driven aircraft does. Partly this is because, despite the design complexities involved, a jet engine is actually a much simpler system.
The idea of not having to haul up a full load of oxidizer on an orbital launcher is a tempting one, but it doesn't come easy. One of the big advantages of a rocket is that it can push up above the bulk of the atmosphere when it's still traveling fairly slowly and do most of its accelerating in a near-vacuum. This reduces aerodynamic drag, aerodynamic heating, and dynamic pressure forces. All of which are some of the most pernicious problems to deal with in a launch vehicle. No few launch vehicles have been lost just as they reach "max-Q" (the moment of maximum dynamic pressure), and for an air breathing launcher it would likely be forced to fly through even more severe aerodynamic regimes than most rockets for significantly longer periods of time. This is hard on the vehicle design, hard on airframe longevity, hard on the thermal protection systems, and hard on the whole vehicle in general.
So on the one hand you have a vehicle which requires significantly more robust engineering and significantly more complex engines and overall design while probably having a shorter total service life. And is perhaps some significant factor riskier to fly in general. And on the other hand you have dead simple basically 60 year old engineering that is just put together sensibly, flown within a familiar flight envelope in a way that minimizes risk and iteratively improved to continuously shave off operating costs. It's a pretty safe bet which one is more likely to actually lead to lower launch costs.
Certainly, but if they pick a flight profile similar to modern rocket powered launch vehicles the advantages of the engine almost entirely evaporate. In order for the engine to be worthwhile the vehicle needs to spend a lot more time in the troposphere and lower stratosphere than any other launch vehicle, and that gives rise to all the problems I described.
Your requirement that they follow a rocket flight path is arbitrary. They'll use the best profile for the technology.
You're also underestimating how hard rockets work while still in the atmosphere. For example, the shuttles SRB work entirely within the troposphere and stratosphere. They're about a million pounds of propellant each, and together they make up 70% of the shuttles lift off weight. If you eliminated the need for the oxidizer in the SRBs, you'd save nearly half the entire weight of the shuttle. Because of the non-linearity of the rocket equation, saving weight produces compounding advantage, so this would be huge.
I think you're misreading what I'm saying, let me see if I can be more clear.
The key goal of an orbital launch vehicle is generating the necessary speed for orbit (over 8,000 m/s, around mach 25). The difficulty of reaching the altitude of low Earth orbit is inconsequential in comparison. A rocket has the advantage that it can do its accelerating wherever it's more convenient, so the typical flight profile is first up and then over, because it's a hell of a lot easier to accelerate and travel at high speeds above most of the atmosphere. For example, the Falcon 9 reaches an altitude of 5km before it even goes supersonic, and will reach an altitude of 30km within the first 2 minutes of launch.
An airbreathing engine however needs to stick around in dense enough atmosphere for its engines to work. And if a vehicle relies on a significant amount of airbreathing then it needs to spend a significant amount of time in that denser atmosphere. And that means that it needs to do more of its accelerating in denser air, which means that it will encounter higher aerodynamic forces, higher drag, more heat issues, a higher max-Q, etc. Those sorts of forces tend to be the "long poles" that aerospace vehicles are designed around, it dictates everything from the materials used to the type of construction to the service life of the vehicle's frame, etc. This is something that positively cannot be avoided for an airbreathing vehicle.
Sure, the SRBs generate a ton of thrust on the Shuttle, but they also help push the Shuttle quickly to higher altitudes and lower air pressure. Before the Shuttle hits mach 2.5 (of 25) it is already at an altitude where atmospheric pressure is 1% of sea level.
As I said before, mass isn't the big driver of cost in orbital launch vehicles, cost comes from complexity which comes from operational complexity (flight profile, staging, etc.) and design complexity (engines, control systems, handling, etc.) A vehicle which saves fuel but increases operational complexity is not a cheap vehicle. Fuel costs around $1,000 a tonne, whereas an engine can easily cost $10,000 / kg.
The biggest win that a vehicle like Skylon would have initially is that it might make it easier to make reusable launchers. If that's the case then even an expensive launcher which can be reused only a handful of times might still be useful in reducing overall launch costs. But if an entirely rocket based vehicle can be made to be reusable then it's very unlikely to have better overall economics or operating characteristics, for all of the reasons I've listed previously.
I think we're mostly in agreement now, just we disagree on our guesses of benefits vs risks.
The key point seems to be the complexity penalty of adding airbreathing to the engine vs the weight savings of less reaction mass. If we're comparing reusable apples to apples, this is really the value proposition. I'm clearly more optimistic on this point.
Also any engine that uses ram effect becomes more efficient at higher speeds. The SR71 uses less fuel per unit of distance the faster it goes, which is a bit counterintuitive. How big a benefit this is for space launch I can't really guesstimate but it's probably minor.
Reaction mass savings means more than just oxidizer material cost though. It ripples through the whole design. There aren't many times when the mass fraction of a rocket is working for you instead of against you.
I think we skipped over that a horizontal takeoff requires a lot less launch infrastructure. But being smart with rockets and launching from a barge in the ocean can equalize things.
As a summary, I think you and Elon may be right about Skylon for space launch. Mass produced rockets can get pretty cheap, and SpaceX does aspire to full reusability.
But space launch is only one of the two applications of a design like Skylon. Nothing SpaceX develops will be used for terrestrial transport. You aren't going to take a rocket to visit your family for the holidays, so Skylon may find a market there.
Skylon also could be used as a WhiteKnight style carrier for a more traditional second stage, which might still be interesting for space launch, but I'm pessimistic on this point because I think if the numbers worked the air force would already be using such systems instead of Deltas.
Skylon can also hedge that their high flow flash chiller is useful in other applications, and apparently they've developed an interesting high temperature composite material.
So on the whole I think it's interesting to watch what happens to them, even if it's not a sure bet.
The cost of fuel and the cost of fuel tanks is an insignificant part of the cost of an orbital launch, around the 1% level. The major drivers of cost are overall system complexity and manufacturing cost of the engines. And here's the big problem for a Skylon spaceplane, rockets are fairly simple systems whereas hypersonic airbreathing engines are extraordinarily complex and difficult. And if you can manage reusability on your launcher then the ordinary rocket engine wins hands down.
The reason why the jet engine won out over the propeller in civil aviation is not because of the higher thrust or better performance of the jet, it's because of lower operational costs. A jet powered aircraft requires less maintenance per passenger-mile than a propeller driven aircraft does. Partly this is because, despite the design complexities involved, a jet engine is actually a much simpler system.
The idea of not having to haul up a full load of oxidizer on an orbital launcher is a tempting one, but it doesn't come easy. One of the big advantages of a rocket is that it can push up above the bulk of the atmosphere when it's still traveling fairly slowly and do most of its accelerating in a near-vacuum. This reduces aerodynamic drag, aerodynamic heating, and dynamic pressure forces. All of which are some of the most pernicious problems to deal with in a launch vehicle. No few launch vehicles have been lost just as they reach "max-Q" (the moment of maximum dynamic pressure), and for an air breathing launcher it would likely be forced to fly through even more severe aerodynamic regimes than most rockets for significantly longer periods of time. This is hard on the vehicle design, hard on airframe longevity, hard on the thermal protection systems, and hard on the whole vehicle in general.
So on the one hand you have a vehicle which requires significantly more robust engineering and significantly more complex engines and overall design while probably having a shorter total service life. And is perhaps some significant factor riskier to fly in general. And on the other hand you have dead simple basically 60 year old engineering that is just put together sensibly, flown within a familiar flight envelope in a way that minimizes risk and iteratively improved to continuously shave off operating costs. It's a pretty safe bet which one is more likely to actually lead to lower launch costs.