There are geothermal plants but from what I remember they aren't that amazing.

We do do geothermal, just not much of it. There are two main categories, power generation and heating/cooling using a heat pump. Power generation requires a power plant, while heat pumps can work extremely efficiently on a building-by-building basis. Heat pumps are great, but can't power a city. For that you need geothermal, which generally means 'hydrothermal' because they all use steam to turn turbines (or to heat other fluids that turn turbines).

Basically, you're right about the limiting factor being that it's expensive to drill deep holes. The heat distribution in the earth's crust is highly variable, so geothermal is more favorable where the crust is hot, because you don't have to drill as deeply (and install and operate equipment at great pressures). In a favorable area, such as near active volcanoes or large hot springs, geothermal may be a viable choice if the area is available for drilling (i.e. not in a national park) and close to a metro area or other high-capacity electrical grid.

When all of these factors work out, geothermal is great. The Geysers north of San Francisco has a capacity of about 725 MW[0], and basically powers SF. It loses ~1% production annually, though, because the hot water powering the plant becomes extracted, and cold water pumped in cools the rocks slightly. Note that the rocks will re-heat themselves, but it takes time.

However, it doesn't always work out. Much of the hot crust in the US is in Nevada and Idaho, where there are very few people, and not a lot of groundwater. Colder crust has the hot water too, but it's at great depths (e.g. 3-4 km) which would cost maybe 50-100 million USD to drill to. (A recent temperature test well--not production--well near me cost $3MM to drill ~1000 feet, and it gets exponentially more expensive with depth. The well, located by a small hot spring not far from Seattle, had a temperature inversion and turned colder with depth, so the money went down the drain).

I do some consulting work for a company[1] that is experimenting with a technique called 'Enhanced Geothermal' or EGS that tries to exploit hot, dry rock by pumping cold water in through an injection well and pumping now-hot water back out at a nearby production well. This opens up a lot more exploitable ground[2] but is still very expensive and not yet proven for production. A test is in progress at Newberry volcano in Oregon[3].

However, I have high hopes for geothermal longer-term. I think that the drilling costs may get a bit lower, especially as a lot of the technology is developed by oil and gas companies (and associated drilling companies that are now or soon to be out of work, with current low oil prices). Additionally, once EGS is more proven and therefore less risky, investors will be willing to put money in. I don't know if it will ever as profitable as fossil fuels, but once the geological and technological uncertainties are lowered, then it should start to become a great, fairly safe investment (i.e. the demand for electricity is unlikely to drop down much).

[0]: http://large.stanford.edu/courses/2011/ph240/halasz1/ [1]: http://altarockenergy.com [2]: http://www.nrel.gov/gis/images/geothermal_resource2009-final... [3]: http://blog.newberrygeothermal.com/