I agree it is a huge and short-sighted mistake, but I can't come up with an example to show it. It jsut seems to be conventional corporate wisdom with little to base it on.
He makes similar points you have that it doesn't make sense to charge new customers more than existing clients because existing clients add more value through things like referrals, they buy more services at lower cost, they no longer incur acquisition costs, etc. An example of trying the alternative would be places like my gym which happens to be a national chain - whose membership services I hate but I bought a while back and they don't up membership costs for old clients but they do for new ones. My membership costs probably about half that of new members because I joined a good 6-7 years ago which makes it rather unlikely that I'll ever give it up.
Personally I think the reliance and dependence of a lot of service companies like Comcast on the value of their infrastructure is a dangerous game given how much the cost of infrastructure has fallen and continues to fall.