If they utilize the waste heat for eg. district heating, the equation looks a lot better. Unfortunately most cities don't have district heating.

There are already lots of sources of low grade waste heat. Since we aren't exploiting them, it must not be worth doing.

So, for nuclear district heating, you'd need higher quality heat, probably steam.

You don't use steam for district heating. There are two problems here, the first one is that district heating doesn't exist in a lot of places so the infrastructure needs to be built. The second problem is that people are afraid of nuclear so running water through their house warmed almost directly with nuclear heat is a sensitive topic. It's not actually dangerous of course, but perception matters. Actually there is a third reason now that I think of it. Nuclear plants are usually not close to big enough cities to absorb that heat, but it's possible to transport the water over 100km while still retaining enough heat for district heating.

Some district heating systems do in fact use steam.

Losing 50%-67% of the energy to waste heat is not the roadblock to new nuclear.

The roadblock is the immense expense of a massive, complicated, intricate machine requiring massive workforces of highly skilled construction labor.

Shifting from heat conversion to some sort of direct conversion, and in the process ideally eliminating a huge amount of the construction expense, is the way out.

As our economies become ever more advanced, skilled labor becomes ever more expensive. Our existing fleet of nuclear reactors is much like the intricate cathedrals of past centuries. We could build in that style, but the expense is much higher today than it was back when the cathedrals were first made.

You mentioned energy efficiency, I simply provided a possible way to boost that efficiency up a bit.

You are right! I must have been thinking of different types of efficiency and written very unclearly. A pet peeve of mine is people talking about "efficiency" in energy and not defining it or switching between definitions sloppily, and I seem to have done it there. In any case apologies for not being consistent.

But why is it so hard. A steam engine in itself is not complex, we had locomotives for a long time, they didn't have high performance welding for example. Why can't we dumb down nuclear reactors? While keeping it safe. Perhaps by using material that won't cause meltdowns or using heat from nuclear waste?

If you have 20,000 welds that each need to last 30 years without being fixed later, that's an order of magnitude difference in quantity than the corresponding 2,000 for a coal boiler, in addition to a few orders of magnitude difference in the necessity of lasting a long time without repair.

It's a hell of a lot more than a steam boiler, even if ultimately that's the goal. The nuclear island is a hell of a beast of complexity, size, and quality.