Fundamentally, SpaceX chose to go with the cheapest approach that works rather than the "best" approach.

Take, for example, the engines. The Falcon rockets are kerolox (kerosene + liquid oxygen) engines. You get substantially more performance from hydrolox (liquid hydrogen + liquid oxygen) engines. The faster you want to go the more difference this makes. However, they recognized that kerolox is far, far easier to work with and that to put a kilogram in orbit on kerolox is cheaper than hydrolox, even though you need a bigger rocket to do so.

I can't find any other western rocket that's anything but hydrolox by the time it reaches orbit. But SpaceX decided that standardizing on the same engine (the upper stage of the Falcon rockets uses the same engine, different engine bell) was a better deal. It's all about mass production. They build a *lot* of Merlins so the cost per engine is a lot lower.

Starship is using the same philosophy. Metholox (liquid methane + liquid oxygen), a bit harder to handle than kerolox but still far more friendly than hydrolox. Same engines throughout. (Although they might have different bells on some of the engines. A bell meant for vacuum works better in vacuum, but at best worse in atmosphere and at worst will be destroyed. The Shuttle OMS engines had a minimum altitude requirement (actually, maximum pressure) and could not be used to help with a bad landing. Light them too low and they would crumple.)

SpaceX make/made things affordable because everything they use is cheap, and have rather focused on an iterative approach. Whereas traditional starships have used expensive materials and astronomical R&D costs, to make sure everything is _perfect_ prior to launch - SpaceX did it the other way. Use good engineering and build the best you can with the cheapest material possible, and start launching stuff ASAP to see where things fail, and then iterate. They also built on existing tech, and saved a bunch there - compared to building something entirely from the scratch, and having to commit to something very early on (even though it might not be optimal in the long run).

With defense you'll have to build stuff that works flawlessly, spend a fortune on design and testing, build products which can be stored for decades (and used decades later), buy parts and materials which meet the specs to endure all that. Economy of scale doesn't really apply to cruise missiles.

But their (Ares) thesis is that they can bring down costs by making smaller but more cruise missiles, which will yield the same force as the fewer but larger cruise missiles.

"With defense you'll have to build stuff that works flawlessly." --> I don't really get this point. In what way is the SpaceX approach different and not flawless? There's never been any more reliable rockets than SpaceX's ever?

And you seem to argue that SpaceX has the advantage of economy of scale (reminder: it's a re-usable rocket they're making) and cruise missiles are somehow less economy of scale? Can you elaborate on what you mean here? I would think that the number of missiles produced is at least a few orders of magnitude more than Falcon 9's?

When you build military equipment, you have to build them to last for decades. 40-50 years, easily. You'll develop a base model, which will then get forked to the various specs that your customers want/need. Differences could be what launch system they're made for(TEL, jets, ships, etc.), what command and control systems they're going to interface against, and numerous other things. Integration is a huge part of these deliveries.

So instead of designing and manufacturing, say 10000 cruise missiles of exact same type, you're in reality manufacturing 1000 for the US military, 500 for the UK military, 500 for the French military, etc. - with every batch going through different specs, testing phases, etc. And that's after you've won the contract, which in itself can take years to land.

Contracts can of course be cancelled if there are too many issues.

If we're going to use the SpaceX analogy - imagine if SpaceX had to first design a base model, but then fork that to say 10 difference versions because 10 different countries / agencies want to purchase their own rocket, and then spend 10 years on each to ensure that they work perfectly to spec for the 10 different spec sheets you've received.

And if it turns out that the cheap material you used to bring down production costs won't cut it, because it is expected to last 15 years instead of 40, then that's a deal-breaker.

> expensive materials and astronomical R&D costs, to make sure everything is _perfect_ prior to launch

I think Boeing have "upgraded" their approach ...

I know that Anduril has a different business model: they completely fund the development of new products, and then sell those products to governments.

The incentives are better aligned to reduce development costs, since Anduril does not earn until products are rolled out.

Maybe this will lead to a cheaper / more efficient defence industry?

The barrier to entry is lack of an existential threat. The USA faces no real threats to the homeland and the military budget, while large, is still affordable. So cost effectiveness and efficiency isn't a major priority. Whereas Ukraine has had to get creative with cheap drones and improvised cruise missiles in order to survive.