The proposition bundles two claims: nuclear is a good way to make carbon-free power, and AI's demand is a good reason to make it. I'd say yes to the first and the thread is mostly arguing the second. If the demand is the reason, then the demand has to be worth it, and nobody here has made that case. They've made the case for the reactor.

Let me make the case for the demand, then, since nobody will: the same grid argument was made against electrifying anything. Nobody needed a fridge at 3am either until everyone had one. If the models turn out to be worth what their builders think, the power will look cheap. If they don't, you've still got a reactor and a grid that can take it. The asymmetry runs toward building.