Nuclear weapons were a major use case. Ballistic missiles wouldn’t use GPS in flight, but they need to know where they were launched from to a high degree of accuracy. That’s easy for ICBMs in silos, but a major problem otherwise. GPS greatly improves the accuracy of SLBMs and mobile ICBMs (which we never actually fielded, but that was the idea).
Wikipedia discusses it, although its source is pretty weak. The Navy’s TRANSIT system, a predecessor to GPS, had Polaris submarines as a major use case.
I’m not sure just what they do, but there are several possibilities for a submerged launch to use GPS. The sub could surface periodically to take GPS readings and keep the fix up to date with other systems, or it could run an antenna up to the surface, or it could let the missiles acquire their own fix after they breach the surface.
Keep in mind that high accuracy is mostly important for a first strike, when you’re trying to destroy the enemy’s nuclear weapons in hardened targets. A second strike can be much less accurate and still be useful. So the idea that you can’t use GPS because it will be destroyed in the war doesn’t necessarily apply.
The missile accelerates so quickly that once you factor in the lag time for obtaining an accurate GPS fix without current ephemeris data it would be effectively useless.