Of course, I doubt it'd come as a surprise to anyone that these payloads are likely insanely expensive as they too have gone through the govt procurement system (which seems to make everything cost 100x what it should cost). At which point it makes sense to avoid losing the payload even at ULA pricing.

$350x1.00 = $350 million for ULA $100x(1/.95) = $105.2 million for SpaceX

The difference is then $245 million or so. So, the government needs to justify why reducing their risk from 5% to 0% on losing a payload is worth $245 million. Purely economically, the payload would need to be worth about $4.9 billion to justify this cost. If we start talking about State secrets, actual production time for the payload on top of the economic cost, redundancy requirements for intelligence gathering, technological advantage over rival powers (China/Russia) that could be lost if another country recovers the wreckage, and the worth of a citizen's/soldier's life on the front line (e.g. - items that are hard to quantify economically) and you can see how easy it is to justify $245 million.

Plus your buddy works at ULA and it would suck if he had to work somewhere else because they went out of business.