Why calculate when we have lots of empirical data? There are plenty of grids around the world are deriving 40% of their energy from wind and/or solar. Up to 65% is relatively easily accommodated according to grid engineering orthodoxy with current typical infrastructure/tech.

Doesn't this mostly happen in counties like Denmark where it's relatively easy to buy / sell energy to the grids of neighboring countries?

This would work well in denser-populated parts of the US, like northeast, and would be harder in Midwest.

HVDC will handle moving energy on continental scales. China makes power in northwest inner parts and moves it thousands of kms to the coast for consumption.

Intermittent energy. Doesn’t work well for manufacturing at all. Your argument, without nuclear, is an argument for natural gas and coal.

Industrials need steady power to function. Without nuclear, the world is going to just use natural gas. Coal for poorer countries. Solar and wind peppered in but not trusted for any base load and regionally limited.

Analysis put covering the whole US with solar to cost about $5 trillion and the world about $65 trillion dollars. A lot, but it would pay for itself in about 5 years. The only blockade is all the people making all these trillions from the oil and all the power sources that make them all this money.

We’re never giving you 5 trillion.

$5T sounds like a lot but it's spread over the lifetime of the panels, 20 years. What do utilities spend on capex and opex over 20 years.

People will loan it and be paid back, with interest.

Who’s “we”, you and the mouse in your pocket? The solar industry is getting that money, one homeowner, building owner, or lot owner at a time. And then, over time, they earn a profit in the generated energy! Sorry you hate making money doing nothing!

We'll sooner build green hydrogen generators and peaker plants than those modular nuclear plants by the looks of it.

If you want to see some big numbers, look at fossil fuel capex. Those big numbers have been spent year after year for decades on getting us into this problem. Don't expect to get out of it for much less, even though individual renewables projects look cheap by comparison. It has to add up to the same power output.

The problem with that is that nobody who is seriously doing energy modelling would propose to do all storage with batteries. That's more a popular misunderstanding than an actual energy concept.

If you plan for a 100% renewable grid what you do is that you use different types of storage for what they're good at. Batteries can play a role in short-term storage, but for seasonal storage they're clearly not a good choice. Electrolyzers and H2 gas peaker plants (or maybe ammonia instead of H2) look like the most promising option.

I wish more people knew this, because all this "I calculated how many batteries you'd need for your renewables, we need nuclear!" clearly is neither an informed take nor helpful.

Let me try some napkin math with nuclear. France is usually taken as a positive example for nuclear. They have 56 reactors on a population of 67 million. For the whole world to get to that level we are talking about 6k-7k in reactors. There are a couple hundred in the world already so let's make it 6k. The costs depend on a lot of things, but 10 billion a pop is a nice round number and for the majority of the world a low estimate. So roughly 60 trillion in construction costs only. Then we need to run them and eventually decommission and store the waste, which even with much smaller scale nuclear programs today will cost hundreds of billions.

Not a realistic solution. We can do more, faster and cheaper with renewables. If you can't get to 100% with renewables, nuclear should anyway be the last option.

Given that the world spends trillions annually on fossil fuels, spending trillions on batteries could still save money.

That's the thesis of the Tesla "master plan 3.0". It claims it will cost $10T to decarbonize the world, but that it would save $13T by doing so.

Tesla is obviously biased, but so is everybody else so might as well use a source where the biases are clear.