The rate at which data can be transmitted is a function of signal-to-noise ratio and bandwidth. Even with a signal significantly below the noise level, data can be transmitted at high speeds if enough spectrum is available.
Orthogonal frequency division multiplexing uses a large number of parallel narrow channels integrating the received signal over a relatively long period of time. Each channel carries only a low data rate but combining a large number of channels allows high data rates.
This form of modulation is very resistant to noise and multipath because of the long integration period of each carrier. If you are driving and a carrier is momentarily interrupted, it does not cause a loss of information because it’s only a very small interruption in a relatively long integration time.
There is a limit to how long this interval can be and how narrow each carrier can be based upon the acceptable latency for an application. For example, a two-way telephone conversation would need a low enough latency that the conversing parties would not notice the delay. But radio or television could easily handle a delay of several seconds without problems.
Other than regulatory limits, the only limit to the number of carriers and total data capacity that can be transmitted is the speed of the digital signal processors and the bandwidth of the radio frequency amplifiers and antenna. Fractal antennas and newer semi-conductors have extended these limits quite a bit.
This is going to continue to expand to provide faster wireless data rates as faster digital signal processors and better algorithms become available.
Single channel per carrier services are going to largely go away and ultra-wide broadband internet will occupy most of the spectrum with radio and television being largely replaced with audio and video over Internet Protocol across ultra-wide broadband networks.
Economies of spectrum, power, and infrastructure will fuel this conversion. A public tired of having half a dozen mega-corporations owning all of the broadcast stations in a market and hungry for choice beyond what satellite operators offer as well as the superior quality that wider bandwidth will make available will also factor in.
Internet automobile receivers are already available. These have the advantage of making a much larger amount of program sources available than even satellite services like Sirius. The horrid licensing arrangements that the various record labels and artists unions impose on Internet broadcasters places some constraints on music being broadcast via this medium, but I think eventually enough artists will go Independent that these organizations will be forced to change their business model or die and become irrelevant. Why should artists get less than 1% of sales and put up with very limited exposure just so labels can get rich?
Internet stations have a world-wide audience by the very nature of their operation, but at the same time they’re competing with a much larger number of stations than conventional broadcasters. This will make very specialized program economically viable and more generalized programming less so. Unfortunately, it also will continue to kill local markets and stations that cater to local markets.