Welcome to the wonderful world of government procurement and R&D. That cost estimate for the CEV doesn't surprise me very much. It's intended to be a modular vehicle capable of performing multiple missions. As such, it's going to require more R&D than a single mission system. Avionics and software development costs alone could easily top a billion dollars. The actual production costs will be much lower.

You mentioned the cost of the Seawolf submarine. According to this source:

The National Defense Authorization Act for FY 1996 preserved the third Seawolf nuclear attack submarine, though it funded SSN-23 at a lower level ($674.5 million) than the White House originally requested ($1.3736 billion). The estimated total cost for the three Seawolf class submarines is $13.1243 billion, or roughly $4.4 billion per submarine. (Source: Department of Defense Selected Acquisition Reports, Program Acquisition Cost Summary, March 31, 1996.)

I don't know if that $13 billion price tag includes the R&D or if that's just production costs. The R&D for a new nuclear attack sub is huge. You're talking about all sorts of new systems, probably including new reactors, sensors, weapons, etc. None of that stuff comes cheap.

As David pointed out, that $15 billion price tag is not too surprising. Airbus is spending somewhere around $10-12 billion to develop their A-380 airliner. That cost may include the cost of tooling and plants to build the new plane, but I'm not certain. I know of military satellite systems like Milstar that cost many billions in R&D and upwards of a billion each for the satellites.

In addition to the costs for developing the CEV, any manned mission to the moon or Mars will also require development of landers, living facilities, rovers, space suits, and a host of other things. Some of these things can use existing technology but a lot of new things will have to be created and tested. Testing these systems will cost buttloads (an imprecise measure, to be sure) of money. Computer simulations and CFD can do a lot, but you don't want your first test of a new heat shield design to be the one a few feet below your ass. A single flight test of a new heat shield can cost many million dollars.