Because even when launching that far from the target, the actual transit journey to the launch point itself and back, may be in airspace which at best is monitored by adversaries or nations that can give adversaries early warning of an attack, or at worst be in airspace that is actively being patrolled and defended by adversary forces (even if one degrades those forces).
You probably don't need edge level VLO to get through there. Pacific, or even Arctic is stupid big, even it's narrow points, and RWR favours attacker. Opponent has only this many points of presense(aircraft, ships) there (if any, it's actually damn far), and it may be simpler to just make those mindful of their misconduct(J-36).
Given lack of dedicated interceptors(and their numbers), i frankly wonder if modern US is truly up to the task to intercept even bear attack with 5000 km weapons. Sector of circle of uncertainty is just immense.
Maximum VLO is quite contradictory to getting good payload to destination. Furthermore, if you actually invested into it, at some point relying on separate penetrating munitions just decreases chances of succesful strikes and possibility of successive attacks, while not increasing bomber survivability
that much.
2000lb jdam and huge IRBM may hit with same warhead, but in one case you waste casing, in another - 10t of precision electronics, high tech SRF(which can only be made in 1-2 places in entire country, and got forbid something happens to that facility).
Finally, survivability is a weighted metric. Simplest place to hit bomber isn't air, it's base - and there, both are equally vulnerable.
But bomber more tilted to use of simpler munitions can be effective from way more places.
Like, B-21 can serve as example here. It god just one bay of B-2, which already wasn't exactly huge.
And all of this weighted against 2030s environment, where we must seriously consider CCAs. Which we can carry ourselves(or, say, refuel from time to time)...if they will fit.
