We are not running out of hydrocarbon fuel sources that quickly. There is still a 73 year supply of cheap oil(cheap meaning able to produce gasoline for less then 4.10 2001USD/gallon assuming a 5% increase in demand compounded yearly) based on known recoverable reserves. And there is a 225-300 year supply of coal, assuming a 4% increase in demand compounded yearly. And a 150-200+ year supply of gas assuming a 4% increase in consumption compounded annually.

My number was based on the current estimated known recoverable reserves from cratonic(on shore) oil drilling only, from various AAPG Bulletins. The peak oil graph you guys use alot on your sites is on the Hubbard Decline Curve. However this only counts proved recoverable reserves rather than known recoverable reserves. Proved means that you have 4 wells within the drainage radius of the proposed well site, which ensures that you will hit oil. Putting this way, proved reserves are for areas with active drilling, while known reserves are estimates based on geology but in sections of oil fields without enough active drilling to prove the reserve exists. I always used the lower estimates though, so the number should be as conservative as the current data allows. Hubbard Decline Curve is also optimized to predict decline of oil production from Texas limestones, in fields where all data is available(Texas law states that all companies have to give their well logs and production reports to the state and that they must be accurate), and that there are outside markets that cause diminishing returns from marginally producing wells. This was an excellent graph that tracked the production of Texan oil very well. However a graph showing the worlds production of petroleum as it is exhausted would be a descending plateau rather than a peak, since there is no outside source of oil to make the marginally producing wells undesirable. So instead of a 150 year rapid decline from a peak, you get a 70 year plateau and a steep drop off at the end.

Even after the exhaustion of all hydrocarbon resources it would still be possible to make hydrogen to fuel internal combustion engines(methane engines are in common use today) using electrolysis. The only thing humanity really needs to continue industrial production is electricity. With enough electricity many things that are not currently economically viable become possible. And with nuclear breeder reactors U238 and Thorium can be breed into the nuclear fuels Pu239 and U237 respectfully. This will provide sufficient electrical generation capacity, assuming an 8.5% increase in the consumption of electricity compounded yearly for population of 10 billion for at least 30,000 years. Supplementing this with solar, wind, and water power will ensure our survival for the foreseeable future.

Of course generating hydrogen through hydrolysis has a negative energy return. The point is that if you make electricity ridiculously cheap, which you can with nuclear breeder reactors, you can produce that hydrogen and sell it as fuel to make a bigger net profit than you would if you produced less electricity and sold it for a higher price per KW/hr. Currently you get back about 12/1 out of gasoline for the energy you expend to find, extract, and refine it from oil. But the average return for a non-breeder light water reactor the energy return is about 55/1(Bulletin of Atomic Scientists, can’t remember the number right now.). With a breeder reactor that return by be as high as 80/1 but lets assume it is 55/1. The best you can expect as a return from hydrogen produced through hydrolysis is about 1/3 if you are using the hydrogen in an internal combustion engine. So the return for hydrogen fuel produced solely by electricity derived from a U238 light water moderated fast breeder reactor would be 18.3/1. Which is still slightly higher than the return from modern petroleum, though not as high as the turn of the century return. Back then all oil wells were open flow(similar to artesian water wells) an gave you a return of over 100/1 since you didn’t have to drill, pump, or frac much.


Sauce:
For the breeder reactors my main source was: Progress and its Sustainability. Choen's Breeder reactors: A renewable energy source, American Journal of Physics, vol. 51, (1), Jan. 1983. Cohen give a good description of the potential of Breeder Reactors. However I must warn you that Cohen is the Edison/Tesla of nuclear power, he is a shameless self-promoter, and half of the time he is talking out his ass. He often words his sentences to play up his own inventions, note that every statement is factual but worded in a way that may be misleading to those unfamiliar with nuclear reactors. Example: Cohen would state that there is a 5 billion year supply of fissionable uranium assuming that all the U238 in the lithosphere was recovered and used in breeder reactors. However given estimated recoverable U238 reserves, an increase in electrical consumption of 10% compounded yearly, and the halflife of U238 there is only a 23,000 year supply. Of course even that assumes that all the 10% yearly increase will come from nuclear energy.