[jewel]

> Internet radio is bandwidth intensive, which tends to eat profits. I would work on something else where you are more likely to make money in the near term.

Out of curiosity, I ran the numbers. 10k users at 96 kbps for two hours a day would consume 315 TB/year.

On AWS, that would cost you $246k/yr, just for the bandwidth. However, if you use a dedicated hosting like 100TB, you can get the cost down quite a bit lower. I think you'd need four of their servers at $215/mo each, which comes to $10k/yr. (Note that a single gigabit port has enough bandwidth to serve nearly all 10k listeners concurrently by itself, so four servers would cover peak usage.)

There will be other costs, of course, such as licensing. The yearly cost is potentially low enough that advertising could cover it. Also, a $5/mo premium account would generate enough revenue to cover another 30-60 free users.

A related thought: Internet radio is an ideal application for peer-to-peer distribution, since even a low-end DSL connection can support several other peers. It can also tolerate dozens of seconds of latency. If each listener is capable of supporting two more listeners on average, you'd be able to run the radio station from your home internet connection without any permanent infrastructure.