[pfdietz]

Recycling of spent thermal reactor fuel doesn't extend things much, since the breeding ratio of today's thermal reactors is much less than 1. CANDU might do a bit better.

The point is that existing nuclear reactors provide only a small fraction of world energy demand. A nuclear powered world has to power everything via nuclear energy, including transportation and industry. To ballpark this, we look at the primary (thermal) energy used by the world, 18 TW, and equate that to thermal energy produced by nuclear (3 GW(th) for a 1 GW(e) power plant.) This would be 6000 1 GW(e) power plants. These would use in excess of 1 million tonnes of natural uranium per year.

[godelski]

I'm asking for the math or a source, because I'm not able to verify your claim and it doesn't match my prior understanding. Especially the recycling comment, as utilization is quite low, allowing for massive amounts of reenrichment. It also doesn't match my understanding of the AP1000 which has a similar uranium consumption rate to CANDU 6. Not burnup, uranium consumption. The thermal energy estimate also seems naive and a bit obtuse as you're also including many things that one typically would not account for when discussing energy production. And again, it also seems naive to pigeon hole the entire world's energy consumption onto a single energy source, as if we did something similar for solar, hydro, oil, etc we'd find massively unsustainable numbers as well.

I want straight numbers and links, because something isn't adding up.

[pfdietz]

Well, let's pull up a first results from googling:

https://www.scientificamerican.com/article/how-long-will-glo...

5.5 million tonne resource, 70,000 tonne/year consumption in current reactors

If you object to the thermal energy estimate I encourage you to do better. You might also take into account that energy demand will naturally increase as the lesser developed world continues its rapid growth (much faster in percentage terms per year than in the West.)