What makes you think they would get no revenue?
The part in the post being responded to where it is suggested that Apple give away mobileMe service for free. Give away free service to millions of people and spend $1 Billion on hardware to support tends not to work out financially.
If you are suggesting that Apple become one of Google's largest AdWords client and paste ads into eveyones mobile me page. Does that really sound like something Jobs is going to approve?????? Wonky Google ads on Apple's carefully manicured web pages.
Without making mobileMe a billboard kind of tough to make the Billion dollars back without resorting to just blatantly selling off folks information.
Google , Yahoo, Microsoft are doing it buy using the "huge datacenters and 'billions' in hardware" to blast ads to people. Apple isn't in the advertising business.
This is likely far more mundane that folks want to admit.
i. Apple has more global customer base. One data center on West Coast and one on East Coast allows them to split the global inquires in half. One side does one hemisphere and the other does the other. Should make folks not in the US happier with response times.
ii. They have a growing push traffic load. Again splitting the work in half makes it scale better.
iii. non-touch screen iPod users shift over to iPhone OS users as the touch drops down to "nano" levels. ( Apple isn't spending all $1 Billion next year. ) 3-4 years from now would be running bigger store and more stuff even if growth slows down a bit.
iv. There is data mining Apple could do just on what folks buy from them once they have even more millions of folks doing that. Find what people want, offer discounts, they buy more , they make more. The majority of that data mining traffic doesn't have to hit the internet... could all happen inside their data center about data that is already there and the "answers" doesn't get shipped out in mass either.
v. App inspections. If developers are going to submit 1000s of apps to Apple each year for approval, some of that will get automated.
vi. back-ups. time capsule in the cloud for a fee.
vii. experimental grid. ( Apple tries out new grid configurations and does testing. So end up have a large scale test system where don't have to leave the building. ) Perhaps when Apple isn't banging on the test grid full strength they firewall it off from the rest of the data center and "rent" it out as virtual machines for other folks who need a test grid.
On likely contributing reason why Apple's roll outs typically suck is that they didn't do reasonable scale testing beforehand. They may in part be due to fact have no room to build a decent dev/test system.
Apple doesn't have to pack all 500,000 sq ft with production systems. If can afford it nice to have space around for unexpected projects that pop up.
(if just leave the section empy of expensive chillers , racks, etc. then can put a internal badminton/basketball court there for fun breaks. Partially joking... the point is don't have to pack the whole building to the gills with gadgets.
Leaving room allow you to do upgrades ... move to new racking/cooling system that is better 3 years from now without having to take stuff outside. )
A similar factor that Microsoft (and others ) are trying is using containers as grid nodes. You ship in a container that has several racks and chilling inside. Plug it in. When need more horsepower get another container and plug it in. When want to swap out old "tech" for much better "tech" unplug container and bring in new one. Moving containers around means you need a big aisles. ( that ends up being "dead" space, from a packing gear to the rafters perspective). This can make the area under the roof bigger, but not really increase the space taken up with computers in the racks.
If Apple is going with the modular container tech then comparing with data center sq footage of 3-4 years ago ... comparing apples and oranges.