Coal is dying but it's still ~20% of US generation. The natural gas share of US generation has gone up.

> Yes, and you use odd bits of land close to consumption sites, many of which will have simultaneous use for other purposes.

Until you run out of those and then your costs increase worse than linearly because you have to start using more expensive land.

> the linearity is an advantage in that it enables mass production, and gets the benefit of the manufacturing learning curve.

Anything you're going to use for a large fraction of the power grid is going to be mass produced.

> Quadrupling your construction costs.

This is the opposite of how economies of scale work. If you make something bigger, the variable costs scale linearly and the fixed costs stay the same but are amortized over more units.

> Also, making your generators much bigger than current practise increases project risk and therfore cost.

This is no different than needing twice as many turbines to generate twice as much power. It's a variable cost, offset by you getting twice as much power without increasing your fixed costs.

> Quadrupling your construction costs again, and decreasing reliability, capacity factor and productive lifetime.

You keep saying "quadrupling your construction costs" without evidence. We build oil platforms in the ocean on a regular basis. They cost some tens of millions of dollars. Existing fission reactors cost some billions of dollars. The difference from being on the ocean is evidently not the dominant cost. And then you don't have to pay for land.

> Seawater is nasty stuff.

Many existing reactors are situated on coastlines and cooled by seawater. It's not some kind of insurmountable problem.

> It will be used for productive farmland, though.

The price of "productive farmland" compared to the price of land near a city is multiple orders of magnitude less, and high density power generation doesn't need that much land.

> Again, why aren't fission or CCGT plants being built in those places? How is fusion different?

New fission reactors largely aren't being built at all because of regulatory suppression. CCGT plants can't afford to spend fuel generating power which is then lost to long distance transmission.

> It's about time to cashflow for utility finance types, and they also tend to want a longer track record than "it worked once".

If it worked once but is now generating ten times more power per unit of investment capital than any of the alternatives then investors would be lining up, and may not even be needed because the plant operator could use revenues from selling such a large amount of electricity to build more plants with.