Being stationary, they can provide reliable communication over a fixed footprint or area on the earth.
That is good news, but there is also bad news. It takes a lot of power to operate radios that can communicate at high speed over a distance of 42,000 kilometers. Increased power requirements increase the weight and size of a satellite, increasing launch cost. (Large antennas and improved radios can reduce power requirements). There is also the problem of signal latency. Round trip transmission time for a signal to and back from a geosynchronous satellite is over a quarter second, regardless of power.
Launch costs and latency are much better for low-earth orbit (LEO) satellites. But, there are always tradeoffs in engineering. LEO satellites orbit the earth in less than 24 hours, so they are not constantly visible from a given spot on the earth. A typical LEO satellite orbit is 90 minutes.
Some low-cost projects in developing nations use LEO satellites, and restrict their access to a few hours of each day.
Another possibility is to launch multiple LEO satellites, each of which contains an IP router. With the proper design, there would always be at least one satellite in view of every point on earth. Packets could be sent up to a visible satellite, relayed between the satellites to one which was a visible above the packet's destination, and downlinked.
A company called Teledesic tried to build such a system, but they went bankrupt before going into operation. Motorola also tried a LEO system for voice communication, but it also failed. However, as radio, power, computing and launch technology improve, such a system may one day be viable.
This would have far-reaching implications for rural, hard to reach areas. It would be particularly important in developing nations, where there is not sufficient demand for Internet service to justify the investment to connect areas outside of major cities.
For more information on LEO satellites, visit the University of Surrey, England, and their commercial spin off.