There is a harsh limit to the incidence angle at the sensor, and this (steradian, as horizontal and vertical angle count) multiplied with the area is a limit fo

There is a harsh limit to the incidence angle at the sensor, and this (steradian, as horizontal and vertical angle count) multiplied with the area is a limit for what focal length/aperture combinations you can reach on the object end of the lens. And yes, lenses increase in cost faster than the sensor area they illuminate, assuming they retain the same wide aperture/large incidence angles at the sensor end. Thankfully, as a photographer you don't have to think about those optical restrictions. It's enough to know that there is a hard limit at f/0.5 that can't be reached with normal lenses (microscope objectives get much closer, though). Practically, everything below f/1.0 is disproportionally expensive. A 50mm/f1.0 medium format lens would be far better than what APS-C could possibly do, just because you'd hit the f/0.5 wall before you'd reach APS-C. The downside of larger sensors is that they are more expensive to make (less fit on a wafer, and the chance of catching a bad pixel is higher), and the optics are inherently heavier and larger, as well as often (but not always, if you can get away with a simpler optical design due to needing a less-open aperture) more expensive. Thankfully, this only matters in two situations: using lenses that reach maximum resolution below f/2.0 with high-density sensors, or operating in darkness/with very short exposures. Both that very-high-resolution scenario, as well as the darkness scenario benefit from larger sensors. The former from better frame-relative resolving power when diffraction limited, the latter from offering wider crop-factor-corrected apertures and thus capturing more light.