As someone that owns a high end imaging studio that shoot "big stuff" every day for a living, we would love it if someone came along and built a sensor using a full 4x5 capture area. It hasn't happened since Better Light and their digital scan backs (and we still use and love our Super 8K-HS), which is now 15 year old technology. The current state of the art is the Phase One 80 megapixel digital back, and Phase One has cornered a very large percentage of the medium format digital market at this point. Both the DSLR and medium format guys are going to start to run up against physical limitations inherent to the sizes of their capture areas, including sensitivity (evidenced by this new Canon's relatively low ISO ratings), and, most importantly, the details of what the optics can resolve. If you want to capture more actual *image information* and not just pixels, you need to scale things up. My guess is that that 50 MP Canon will just make it that much more obvious how soft most lenses are at the corners.
To build what you describe, someone would have to pay for the R & D to design and fab a chip that large. That's no trivial effort. Only a few foundries in the world right now are even capable of producing it. I believe Phase One is using Sony silicon in their latest iterations; previously, they had used DALSA for awhile. Yield per wafer is obviously way lower too with a monster sensor -- maybe two at the most? These guys look at the potential market and have to cost-justify it that way, and the market is just too damn small, unfortunately. If the selling price is six figures (and I don't see how it couldn't be, given that a new Phase One IQ280 is $50K), how many potential buyers are really out there? I just don't see it happening unless someone like the Federal government steps in and commissions their own cameras, like it did with Hasselblad back in the 60s for the space program. With the NSA and Homeland Security's "needs," you never know.