Some years ago, the Deutsche Bahn found out that having a small army of surveillance drones in the sheds would have costed them less than making up for the damage caused by vandals and spraypainters to trains. So defensive uses of drones look like a really cheap alternative.

About using drones vs drones, that would likely end up with dedicated drone classes, like interceptors. Not that it would be too strange: just compare the Global Hawk with the Predator with the Scan Eagle. But the main reason for developing such a class of drones would probably be low costs, enabling your enemy to strike at you with more than you can handle without using drone technology yourself. The question is whether drones really would be the cheapest alternative, when it comes to interception.

As for whether or not they will be good and made in the millions, I think that nothing is impossible, but there need be such a demand. And there's also the fact that prices for military-grade long-range weapons are very high, and cheap gets relative. I mean, cheap is a combination of absolute cost + GDP of the maker/buyer + value of items destroyed or captured or denied to the enemy via the product + GDP of the enemy + human costs that have been avoided + collateral costs/gains (like fighting a shorter war, that should be a net gain).

So I can see millions of recon drones, but I can't see millions of air force drones. I could see a multi-role infiltrator drone for each squad, though. That whole "enter building, scan around, fly a bit high, defend yourself" part would be nice. And some things, that now soldiers have to carry themselves, could be put on the drone. Like radio jammers to be safe from IEDs, or medi kits.