The primary cost for nuclear is the initial construction. An operating plant has the wear items replaced on a regular basis as part of normal maintenance, so it's more like Ship of Theseus than something that has a specific lifetime.

>The primary cost for nuclear is the initial construction.

That's a common claim but it's not really correct.

Firstly, the average plant in the US spends it's construction costs every 2-3 years, I assume the situation is similar in the EU. New power plants are much more expensive of course, because the old ones were built before Chernobyl, Fukushima, 9/11 and other incidents that revealed horrible security flaws in existing plants and procedures.

Secondly, the costs of handling Chernobyl are way, way higher than the total cost of construction for all nuclear plants in the EU combined. Similarly the cost of managing Fukushima is way higher than the cost of constructing all the nuclear plants in Japan. These costs are of course never added to the cost of nuclear power, since nobody operating nuclear plants has any chance of paying such sums. But they are in fact a larger factor than the construction costs were.

These experiences did have some effect, thankfully, in that made new power plants more expensive, because we know that we must take any conceivable measure to prevent more Chernobyls and Fukushimas from happening.

And thirdly, the reason why Sweden closed the two most recent reactors was cost, despite them being paid off a long time ago. Meaning there is significant cost even after the construction is paid.

So it seems that the claim that "construction is the main cost" comes from the operator's point of view. Basically "the construction is the largest part WE have to pay, because taxpayers cover the rest".

That works for things like auxiliary pumps, but replacing the reactor vessel or control room is going to be a completely different game. And all of that is made even worse because of the whole "nuclear" part, with plenty of critical parts becoming irradiated and essentially impossible to touch unless you let it rest for a couple of decades. On top of that it locks you in to an old reactor design, which might be very undesirable for safety reasons.

At a certain point it just becomes cheaper to build a brand-new one right next to it and decommission the old one, even with a nuclear reactor.

Hopefully, small modular reactors (SMRs) bring down construction costs and build times significantly. The economics would be better than 2010s fossil fuel prices if companies like NuScale and TerraPower deliver on their ambitious, but attainable promises.