An AT&T excecutive said that due to surge in online content, expecially video, the Internet will hit full capacity by 2010. According to him, $130 billion is needed to improve the global Internet infrastructure.
Certainly, the Internet is not free. The marginal cost argument ("as the marginal cost approaches zero, the product will naturally become free") misses a basic truth in business and economics.
Let's assume the marginal cost of operation is zero, as pro-free people seem to argue (which can never be true by the way). Even at the zero marginal cost, you cannot price the product zero, because then you cannot recover your investment.
A roller coster in a theme park requires a large investment to build, but once it is built the marginal cost of running it another round is almost zero. So, the theme park should take customers for free? Of course not. (If you can make money elsewhere, you might do this. This is another area where pro-freers are misleading, which I want to talk about some time.)
The Internet economics cannot be understood correctly if you only look at the marginal cost during the operation stage. Whether you are an ISP, a web service, or a content provider, your cost is concentrated in the investment and fixed costs as opposed to the marginal costs. And when you were deciding to invest in it or not, you certainly assumed some revenue. Don't forget that.