Missiles are at least an order of magnitude more expensive than the most expensive cannon fired Kinetic energy penetrators.
If you select missiles as your main armament and the number of stowed kills stays the same, gun designers
are going to be able to show that they can apply all of the exotic weight reduction techniques that were previously
regarded as cost prohibitive.
The question is how many LOS "anti-tank" stowed kills is needed? It should also be noted that historical ammo count is not representative: accuracy was poor and pk was low per shot. When pk is high and shots are not used for suppressive effects, ammo needs are low.
The problem with using a large, really high velocity gun is that your stowed kills against other targets is reduced. There is orders of magnitudes more infantry, more mini-ugv, micro-uav, holes in the ground, light/medium vehicles that needs servicing.
The previous conflicts in the past few decades have shown extremely few tank on tank engagements, and none that had a strategic impact.
-----------------
Even if we assume a conflict where formation tank on tank battle occurs, how likely is it we need large number of LOS stowed kills? Should we commonly expect 5+ kills per rearm against against a opponent that needs 130mm+ HV guns to defeat? As the life expectancy of the combat vehicle goes down the amount of stored kill needed also goes down since it is unlikely to fire them all.
You need a opponent that:
1. Manage to penetrate USAF fires
2. Manage to penetrate army BLOS fires
3. Arrive at front in dense, numerically superior force
4. With the American tank force not neutralized by enemy long range fires
5. That has combat power that defeats Abrams by protection but is utterly ineffective against FMBT in LOS combat, taking multiple losses per FMBT
6. Yet persist in engaging as opposed to disengaging
7. Under conditions where the American forces can not just disengage as well
8. While being the highly significant point of conflict
The problem is that #1, #2 is extremely hard, while matching FMBT on #5 is fairly easy as it is detected when it fires and joint fires or missiles (of sufficient capability) can defeat it, even if gun development is lagging. Even if technology is unavailable, Brute force methods like huge Kinetic kill missiles or large missiles salvos is still more economic than horrible exchange ratios, and it is not really conceivable way that a FMBT can be defended against all those methods and still pack a big gun and ammo and have mobility. It is also unlikely that sensor outmatch could enable large kill counts as LOS is limited and commercial thermals is getting good enough.
In any case even if this scenario is not literally impossible it is still likely to be very implausible, and optimizing for it as opposed to all the other very common scenarios is wasteful.
-------------------
Though thinking about the constraints a bit, I think huge number cheap stowed kills can enable completely different engagement logic. One can start firing semi-blind into smoke, cover, and poorly localized radar signatures and likes. However with a big chemical gun is still too constrained by ammo count and such such tactic would probably have to wait till practical railguns. You probably also need an anti-sensor revolution for this to be actually useful.