Realistic fusion (with the best understood technology): build powerful magnets around a donut shaped chamber, which allows to contain a plasma comprised of Deuterium and Tritium (both Hydrogen isotops) which is then heated by externals sources. Reach very high temperatures such that fusion reactions occur frequently. Some of this energy stays inside the plasma, and some of it escapes under the form of neutrons. Capture these energetic neutrons in a blanket around the chamber, creating fuel (tritium) and heating water pipes that then drive a normal steam turbine. Tritium is radioactive (but has a very short shell life; just wait a couple of decades), and the chamber may be slightly radioactive after decades of neutron bombardment. There are no problems of long term radioactive waste, and the reactor can't do a chain-reaction, so no Fukushima or Tchernobyl.
I need to explain what Q is in the context of fusion. Basically, you heat the plasma with some energy (Energy In), and the fusion reactions produces some energy (Energy out). Q is basically the ratio (Energy out)/(Energy In). When Q is bigger than 1, we call it break-even. However, (Energy In) is not the actual cost of energy you need to run the whole facility, it is only the Energy that reaches the plasma. The same goes for (Energy out): this energy cannot be captured 100% efficiently. Some of it will heat the plasma itself, some of it will escape but the conversion back to electricity is not 100% efficient.
So in a sense, Q > 1, aka break-even, does not mean commercial fusion, it is only a kind of a psychological barrier to achieve (so this is what the NIF announced; still a major breakthrough). We need at least to achieve (Total Electrical Energy out)/(Total Electrical Energy In) > 1 to achieve commercial fusion. But physicists consider the rest as engineering problems, not physics problems. And great news, there is no theoretical limit on how big Q can be: for example, the sun has a Q of infinity, as there is no required energy input. Current estimates put Q at least 30-40 to achieve commercial fusion (again: there is no physical limit to achieve that, only engineering difficulties).
Main costs are: difficult to define, because we haven't commercialized a reactor yet. I would say, for now, everything around it is expensive (magnets, the blanket, the fuel (tritium)). However, once we have sufficiently understood the optimal parameters on how to produce net gain energy, there is no reason why the design of the reactor can't then be simplified to be mass-produced.
Note: the technology used by the NIF is very different from what I described for a realistic fusion device: what I described is called magnetic confinement, and what the NIF did is called inertial confinement.