Nuclear power has always been a lot more popular on K Street than on Wall Street. At least these sort of overruns pale in comparison to some of the ones in Europe - one in the UK has now become the second most expensive thing ever made by man (after the International Space Station). Lots of nuclear plants on that list, too. One in Finland is now a decade overdue and commercial operation still isn't expected until 2018 - assuming there's not even more delays.
One of nuclear's biggest problems is, it doesn't work very well small. There are some "smallish" modular reactor designs, but as a general rule, nuclear plants are very large structures. Which means, you're not making a lot of them. Which means you don't retire the risk (both financial and safety) very quickly. Nuclear inherently contains a lot of both of those. It can take decades to learn what problems are. And when we redesign systems to start over with a new "generation" of nuclear power plants, that "ironing out the financial and safety kinks" process starts over.
It's unfortunate, but the very nature of fission means going through every element on the periodic table except the extremely short-lived/superheavy ones. Which automatically means facing very significant corrosion and containment challenges. The very nature of a high neutron flux means degradation on its own. The very nature of having exceedingly toxic materials means that you can't allow even tiny amounts to escape, and have to go to extreme levels to prevent serious problems like fires - and not only is your fuel source challenging from a chemical and materials standpoint, but it also can't be shut down quickly. Criticality can be, but the daughter product decays keep the core hot for a considerable length of time.
Nuclear is eminently doable from a technological standpoint. But like rocketry, a lot of things conspire to make it very difficult to do affordably and safely.