The law of diminishing returns suggests that it's unlikely.

The law of diminishing returns has thus far been wrong on battery and renewables. Previous estimates often were wrong due to assuming a slowdown. Not to say it can't or won't happen, but unlikely isn't the word I'd use.

The last two years were weird, especially regarding batteries due to prices increasing, but I'd put that on Covid and the resulting mess.

Especially as solar & wind is already cheaper than everything else by far, so the critical part is batteries which is a broad field with a huge increase in relevance & research over the past few years.

The chances are pretty good, but of course a slowdown could still happen.

Chemical batteries are nowhere near suitable for providing sustained load to the grid, and never will be.

What's happened is they're now cheap enough that for fast dispatch load levelling applications, you can install them where instead you'd have to consider bringing a whole powerplant online inefficiently.

In the next 100 years, I predict no overnight load is going to be handled by installed grid batteries as part of a major energy grid: it'll only ever be backup capacity to make smooth the small scale variations of renewables (i.e. a cloud blowing over a solar farm briefly).

Right now, household batteries are being installed by those that can afford them (and the prices dropping). They are enough to handle overnight load. And electric vehicles are becoming popular, usually with double or more capacity, which will replace the household batteries. The uptake is high and will only get higher because it saves people money. Even people who can't afford a household batteries themselves will be able to get one installed, as the cost saving will allow someone else to profit. So grid batteries are easily able to provide sustained load to the grid, by simply pricing power from the batteries higher than power direct from the generator. Household batteries kick in because it saves people money, and turns out you need very little grid storage at all to handle overnight load. The trick is handling black swan events, such as extended periods without wind or ash/dust blocking sun, across the entire grid area (usually solved by making the grid bigger, such as connecting to neighboring countries).

Any civilisation that can produce a 60kWh battery or three per residence for gigantic EVs can trivially produce a 30kWh battery for overnight storage. Prices are at around $150/kWh at the cell level and assembly costs are plummeting, there's no real reason to think 12 hour Sodium-ion, Iron flow or ZnBR storage won't be available for prices close to the PV required to drive it daily.

It won't be needed because renewables can provide 70-95% of hours just by adding dispatchable loads even without adding more PHES or CSP, but it's not because batteries will be exorbitant compared to non-renewable generation options.

While the law is there, the grid-scale battery tech is pretty far from the point where the returns start diminishing. I'd say that it's now in the lower part of the S-curve, with a lot of lower-hanging fruit to crop.

I'm saying this as a definite proponent of nuclear power and of renewables.

The experience curve for PV follows a power law, where a doubling of cumulative production reduces cost by about 20%. At that rate, when the world is powered by PV, the PV would be producing energy at < $0.01/kWh, and even less in the best locations.

With that cheap solar, it would be better to just heat up a big pile of rocks with resistive heat and use that to drive steam turbines, instead of using nuclear fission to make the steam. Artificial geothermal.