The other reason is there's a lag of several years between when a capability is established (Falcon Heavy) and when large customers can take advantage of it. For example, we really shouldn't expect Starship to make a dent in way satellites and telescopes are designed and built until the 2030s.
DoD and USG learned through the 90s-now how screwy things get when you don't have enough competition or you can't spread your money evenly amongst vendors. They will spend the money now and indefinitely to assure they can get to space when they want to, how they want to.
That doesn't mean they're happy about it though. I'm sure they'd love if everyone was as affordable as SpaceX.
(This isn't an anti-SpaceX thing, really: the dual-provider system (Atlas/Delta) long predates them).
https://en.wikipedia.org/wiki/National_Security_Space_Launch
(And besides, these satellites are orders of magnitude more expensive than their launch vehicle! The NRO's budget alone is around $10 billion per year).
As an example when the UK built the first Astute class submarine about ten years ago they ran into a huge number of delays and technical issues because everyone who had been involved in their previous nuclear submarine construction program in the early to mid 1980s had retired or moved on to new things.