[_delirium]

I would guess your chances of doing #2 are not much higher after buying the photos than they are right now, unless the photos provide something in particular that's interesting that you can exploit. There are a large number of freely-licensed photographs on the web that you can use free right now: everything from NASA photos to things people on flickr tag as cc-by or cc-by-sa, to a ton of stuff contributed to Wikimedia Commons. You could take those, stick them on a website somewhere, and slap ads on them (and people do that). What will 10 million new photographs add? Unique content that isn't already republished elsewhere, of course, but how valuable is that to the image-farm-with-ads endeavor? Will users even know that your images are unique rather than copied from Wikipedia/Flickr/wherever? My guess is that most won't, at least if you display them as a jumbled aggregate set of images: a random unsorted 1 million images from one place versus a random unsorted 1 million images from another place.

It's possible there's interesting subsets buried in there that do 1) provide unique content not already available on the internet, and 2) that you would be able to drive traffic to. I suspect it would be a lot of labor to sort and figure out if that's true, though.