I came to the conclusion about six months ago that the current generation of commonly used wireless gear (Motorola Canopy and 802.11a/g based gear) is already providing equivalent or better speeds than the DSL that is generally available in the US.
You are going to see a lot of higher speed DSL advertised, but it is not often delivered if you have a loop over 4000-5000 feet or so. I used to resell some Qwest DSL and have done a fair amount of troubleshooting on it. What is interesting, is that when you get into the nuts and bolts of troubleshooting a DSL modem, you end up with db readings for signal and noise, just like you see on a wireless deployment. Based on the readings I saw during troubleshooting, I would take unlicensed wireless over the majority of DSL loops any day. Most copper plant, especially in rural areas but sometimes in urban areas as well, is not in good shape and it is not being updated. The copper plant is slowly deteriorating and the telcos cannot build out fiber fast enough to counteract the deterioration – and they have no incentive to upgrade the copper now because they are going to be replacing it relatively soon anyway.
This only takes the physical plant into consideration. Most DSL providers (including the big guys) are oversubscribed on their DSLAMs and Infrastructure points, creating another set of bottlenecks. It is a lot easier to upgrade a fixed wireless plant to higher speeds than copper plant, so wireless has a big advantage there also.
PS – Anyone who doesn’t get the “jumped the shark” reference – here you go.