To stay ahead in the race to the bottom, Amazon will need to continue subsidizing tablets. So when, say, Asus and Google are selling a co-branded $149 tablet, Amazon might sell a comparable device for $99. When the Asus-Google tablet drops to $99, Amazon would need to sell a comparable one for $69 and so on.

It's a losing proposition.

The lower prices go, the broader the audience. You get a lot more people who are unwilling or unable to spend a lot of money on things.

People who won't buy a tablet until it's really, really cheap aren't going to spend freely on Amazon.com. But Amazon still has to lose money on the tablet in order to undercut Google's dozens or hundreds of tablet-making partners.

Any discount retailer, from Wal-Mart to Costco, has to make it up on volume. But Amazon can't. If Amazon doubles the number of tablets it sells, it doubles the amount of money it loses on those tablets. Yet the new customers acquired through those losses are less likely to buy a lot of stuff on Amazon. They're either broke or cheapskates -- that's why they're buying the cheapest tablet they can find. These are not exactly the ideal Amazon.com customers.