Their greatest party trick, is that the contaminated part is basically a self contained "box" that does not need any maintenance for 25 years. and is the only part that needs to be recycled or sent down a mineshaft after use..
Makes civilian reactors with their tons of waste per year and their 40/50 year decommissioning of thousands of tons of steel and concrete seem a bit prehistoric...
Orders of magnitude lower output though. Plenty of accidents too, just better covered up than the civilian ones.
But you could do a modular design. so if a core fails, just shut down that one and distribute the load.
These cores tend to be quite reliable, but the habit of crashing into undersea mountains, combat maneuvering and carrying a bunch load of stuff that goes "blam" does push up the incident rate.
In any case, I would like to see this as a transition tech to fully sustainable energy (meaning large scale storage) and/or a far better nuclear tech, based on either U238, Thorium or whatever is better/safer/less polluting than U235...
A a side note, on ship engines, we are pushing the limits of thermodynamics, right now the big thing is marginal gains, but one you have pushed temps down to ambient, the rest is not reusable...
We manage to separate oil in minute quantities out of waste water to get the last drop into the injectors, but that tech has reached maturity (should be done on land, all that oil in household waste water...).
Hybridization is not useful in our constant RPM/load method of use... And you would need to get a hold full of Li-ION to get trans Channel range...
We use up to 4 different engines sizes to always have the right one for a task...
Most of our engines architectures are 30 to 40 years ahead of what is on the road today.
So you could argue that my industry is ready for the next big thing, ready for a tech change.
But it wont be windmills (silly) or solar (even if 60% efficient at all temp ranges - including nights).