So how small a reactor? And what sort of cost, what sort of setup time, what sort of safety constraints (ie, current reactors are generally water-cooled and need access to large bodies of water)? It's possible to trim the cost of reactors from the billions to the millions, maybe, but how many millions?

Certainly, solar with batteries is a lot more expensive - but to my point, startup cost is low. Latency, or bandwidth? The long-term cost/kwh may be higher for solar, but the short-term startup expense will be much lower, unless you can get the factory-built reactors down into the thousands rather than millions of dollars.

If nuclear is cheaper than solar in a decade or two, but costs ten times as much in the short run, there's a tremendous advantage to solar, battery cost or not. Opportunity cost matters tremendously.