Actually, I was referring to the cost of using a microprocessor and associated ICs to produce an RF output, especially at any useable power levels.
Think of it as attempting to generate, transmit, and receive RF signals using a motherboard and perhaps some logic chips.
Much cheaper to use analog components and simple circuits or simply purchase a prebuilt unit and customize it for your needs.
Moore’s law is strictly microprocesser density and I was thinking of economic scaling, so that wasn’t really an appropriate example on my part. What I meant, though, is that if purely digital radio becomes advantageous commercially, then the cost of making it will dramatically fall. Even though cheap digital components can be used now, they’re still more expensive than analog, as you pointed out. But in time? We’re talking billions of applications here, which would definitely result in a lot of specific hardware creations to support it - LSICs, etc. - which could be churned out extremely cheaply if there is a need.