Canon's point of view is that they have a ton of 500nm fabs. These are extremely old fabs working on ancient processes, but they've been fully paid for, are relatively cheap, and produce sensors at high yield.
It is also the reason Canon can't compete with Sony with regards to sensors; most of Nikon's high DR sensors are produced on a 180nm process, allowing for more pixels as well as a more sophisticated / advanced A2D converter architecture that allows for lower read noise.
It's not a matter of design choice; Canon literally could not make a D800E if they tried: they simply do not have the technology.
Now, Canon currently IS working on its own 180nm technology, but first, Sony has already begun to move to 90nm for their P&S / Smartphone sensors, and second Canon only uses its 180nm right now for its own P&S, instead of larger format sensors.
So on a technology basis, what is correct for Canon to do? On one hand, it needs to chase Sony with higher megapixel and finer-architectured equipment; it's working on that, but on the other hand, it has all this technology and capital invested in its high-yield / low-cost 500nm production process.
What it can do with its 500nm fabs is to move them to medium format. The fabs are now essentially free; they've paid off themselves through years of arduous work producing APS-C and FF sensors. The fabs have extremely reliable technology; manufacturing experience has allowed them to drastically reduce the flaw rate in both APS-C and FF; they probably make more profit per FF or APS-C camera produced than Sony/Nikon, since Sony/Nikon needs to use a more advanced and thus less mature profit. It stands to reason that Canon has a reasonable shot at cheaply producing Medium Format sensors at a price none of the other producers can compete at, with better technology and higher quality than the existing producers.
With regards to the medium-format industry, Canon would likely be able to dominate the industry as its technology seems to be more advanced / mature than existing medium format technology. If you look at existing medium format cameras, many of them, while having far superior resolution to Nikon and other FF devices, they are lacking in high-iso performance and dynamic range. Simply by moving Canon's technology in, you would have something that is drastically superior to existing medium-formats.
On the other hand, by entering the medium-format market with an affordable but extremely capable 3k camera, Canon would then take the performance crown from Nikon. A Canon medium-format would automatically be the best-performing camera in the industry, and the halo effect from having Canon super-medium formats would aid sales in the rest of Canon's industry.
The best part? Sony and Nikon can't chase Canon into the medium-format market. With the 180nm technology, the likely flaw rate will be excessive, so they can't economically compete with Canon in MF. They might be able to produce a MF camera that outperforms Canon's variant due to its superior read-noise at low ISOs, but that MF would likely cost at least twice as much as Canon's offering.
As for the bodies; if Canon opts to buy Phase-One or some other producer, it can easily inherit that company's existing line of lenses. If the body system it supports has a longer flange distance than the EOS system, then that's great; because then Canon can use .5x wide-converters mounted between the lens mount and the EOS lens to adapt the lens for the MF camera.
It won't necessarily be that easy; afaik, you need a different wide converter for each lens, or otherwise need an adjustable wide converter, and there will be image quality degradation as a result, but a Canon MF system would then be supplied with a full range of lenses.