My other concern is that such a complex and interconnected system is likely to be fragile in unexpected ways.
And none of this addresses the land area required for base load solar.
I second that.
The robustness of the current global energy supply system based largely upon solid, liquid and gaseous fuels stems from its flexibility and inherent buffering capability and the fact that lots of the connections/links are loose/weak/interchangeable. (Supplier of fuel xy can be chosen relatively spontaneously, completely unlike some cable which invariably connects two fixed geographic locations.)
An intercontinental electrical grid is in comparison dramatically less flexible. I'd call it grossly negligent to even attempt to implement a world spanning grid exceeding lets say 15 % of total power transmission.
And the overbuilding necessary is usually more in range of 10x assuming favourable storage conditions (which aren't necessarily available).
Mind you, the model misses things like "Bavaria is NIMBYing all necessary transit work" etc. which is why you have weird situation of Germany paying Netherlands to curtail wind farms while buying nuclear from France, because they have to route around NIMBYs.
Only slightly related - I'd love if the cost comparisons in general discourse stopped avoiding the issue of intermittency (that's how you always get such low prices on wind and solar) and used costs that include capacity factor and dispatchability (which https://model.energy/ does). Having funding (and purchasing priority) decided on dispatchable green sources (whether hydro, wind/solar/battery/hydrogen mixes, or nuclear) would probably help a lot in stabilisation of grid and long term decarbonisation, even if it would reduce some frankly pathological incentives for renewable build out (pathological because of linked carbon-producing plant build out to stabilise the setup, while simultaneously using cheap peak production prices to push out stable setups)
The site's model also does not include pumped hydro, or intermediate term thermal storage.
All these extra options can only help reduce the cost of integrating renewables, and make renewables more competitive vs. nuclear.
Isn't the problem that electrolyzers mostly have to run 24/7 because they would otherwise not be profitable? Then all these processes that rely on hydrogen wouldn't actually belong to the dispatchable demand category.
Dispatchable demand could be found in heat pumps (in combination with buffer tanks) for heating homes. (As long as there's sufficient grid capacity)
But using curtailed intermittent power is different: here, the electricity is cheap, and capital cost of the electrolyzer becomes more important than efficiency. Under these different design tradeoffs, one might use a different technology (alkaline electrolyzers without platinium group catalysts, say, rather than PEM cells with expensive Nafion membranes and platinum.)
I have seen claims that alkaline electrolyzers are available for < $200/kW in China.
As for curtailment-reducing tricks, that's a good argument but IMO not changing the base calculus much - the model is about replacing the base demand power sources, so that we can push the extra into peak. Bigger problem is that scaling down demand is problematic for dealing with underproduction, and also simply missing on power (Thus my preference for billing for dispatchable power generation, then extra peak could be funneled easier into bonus things like green hydrogen for non-storage etc.)
On-river pumped hydro is not easily expanded. But there is enormous potential for off-river pumped hydro.