It usually comes down to two factors.
One, the time horizon over which we observe and discount the levelised cost of energy (unsubsidised) relative to the plant’s lifetime. Nukes’ long lives mean their tremendous capital costs’ amortisation matters. In this, the anti-nuke tends to fudge its numbers.
Two, the degree to which we believe the Nth model will be cheaper than the 1st. We’ve built two AP1000s in America, both in the same series. China has built four, one in just nine years (multi-year delay for coolant pumps), in two roughly-parallel series. So the anti-nukes’ too-slow argument is bunk. But the pro-nuke team needs a learning curve barely demonstrated in another country to continue.
The meta-debate, however, is between and and or. I’ve seen no good reason not to pursue both building nukes, where the cost curve is unknown, and building wind and solar, where the details around storage (and how we’ll pay for it) is unknown.