Until someone figures out a way to manufacture antimatter, fusion is by far the cleanest and most abundant source of power we can hope to harvest. We've known this for a long time, but fusion is hard, and it's expensive to build the giant lasers or toroidal plasma containment systems that are needed to get it to work. By most estimates, we're something like 40 years away from an operational fusion power plant.
"Most estimates" do not, apparently, include research being done at Lockheed Martin's secretive advanced development center, Skunk Works. At Google's Solve For X, Charles Chase describes what his team has been working on: a trailer-sized fusion power plant that turns cheap and plentiful hydrogen (deuterium and tritium) into helium plus enough energy to power a small city. It's safe, it's clean, and Lockheed is promising an operational unit by 2017 with assembly line production to follow, enabling everything from unlimited fresh water to engines that take spacecraft to Mars in one month instead of six.
Lockheed's fusion power plant uses radio energy to heat deuterium gas inside tightly controlled magnetic fields, creating a very high temperature plasma that's much more stable and well confined than you'd find in something like a tokamak.
Chase didn't give a whole lot more technical detail, but he seemed confident in predicting a 100mW prototype by 2017, with commercial 100mW systems available by 2022, implying that all global energy demands will be able to be met by fusion power by about 2045. No more oil, no more coal, no more nuclear, and not even any solar or wind or hydro will be necessary (unless you're into that sort of thing): fusion has the potential to produce as much affordable clean power as we'll ever need, for the entire world. That's wild, and we may see it happen in less than a decade. That is, if Lockheed Martin's plans come to fruition, which we certainly hope they do.
Watch the Solve For X presentation video below.