https://web.archive.org/web/20130513055956/http://www.iisd.o...

Which in turn references this:

https://www.nirs.org/wp-content/uploads/neconomics/weltstatu...

Which basically says they have crappy data but then proceeds to lump in a lot of historical costs, like the sum of all taxpayer-funded fission R&D, including things like weapons and nuclear submarines that have little to do with civilian power generation, "liability caps" even though there hasn't actually been a reactor incident in the US that reached the liability cap (and fossil fuels and everything else also has de facto liability caps because of corporate limited liability), and "stranded cost" rules in deregulated energy markets that also apply to fossil fuel generating plants but aren't included there because the numbers are coming from completely different sources. Moreover, even some of the legitimate costs (like the R&D actually used for civilian power generation) are fixed costs that stay the same, and thus decline per kWh, if you build more plants. And that's what I could find in a few minutes of glancing at it.

This is par for the course in these comparisons. People want nuclear to look bad, so they lump in every cost they can find to the calculation, even ones that don't even apply, without doing the same for competing generation methods.

And this is before we even count the massive subsidy fossil fuels get from carbon not being priced.