[_bin_]

Here are Lazard's 2024 LCOE numbers, a popular reference (though they're often too generous regarding renewables). https://www.lazard.com/media/xemfey0k/lazards-lcoeplus-june-...

You will notice that utility PV and CCGT are relatively similar in cost. Of course, replacing most of our energy infrastructure would have massive capex that one would add to the solar option. Note that the solar numbers do not include the cost of storage. And the storage requirements as you replace each GW of generation get higher, not lower.

[kragen]

Unfortunately this PDF doesn't explain Lazard's calculations, so it won't convince anyone who employs any critical thinking. Key factors like types of PV panels, whether tariffs are included or excluded, to what extent the panels' peak capacity exceeds the inverters' capacity (if at all), interest rates, etc., are not provided at all. Capacity factor is only mentioned as a broad range, with no mention of factors like what country it's in, fixed vs. sun-tracking, transmission costs for especially sunny sites, etc. Maybe it's fine if you want a rough guide, if it's correct, but it presents no evidence to show that it is correct.

Some evidence that it's not correct is that Saudi Arabia signed a PPA for wind last year at US$15.65 per MWh https://renewablesnow.com/news/saudi-arabia-signs-1-100-mw-o..., which is barely over half of Lazard's minimum LCOE for wind of US$27. So I don't think they're being too generous regarding renewables, though their solar numbers do look more reasonable.

Here's a quick LCoE exercise. Unlike Lazard, I show my work, so that, if it's wrong, anyone can see that and correct it. Suppose you're a utility that can freely import Chinese solar panels at the current €0.110/Wp (https://www.solarserver.de/photovoltaik-preis-pv-modul-preis...) in a region where single-axis trackers give you a 25% capacity factor, you are borrowing money at 5% to build a 100-megawatt-peak plant, the panel cost is 30% of the total plant cost, and the plant life is 20 years. (These are fairly typical numbers.)

The panels will degrade over the 20-year period, so you'll only get about 90% of your rated peak power on average over that time, and the capacity factor of 25% brings that down to 23 megawatts on average. (This is front-loaded, so the panels produce more toward the beginning of the plant's life when it matters more, but let's ignore that.) This works out to about 200'000 MWh per year.

To amortize a €36.78 million loan over 20 years at 5% you need to pay €2.9 million per year.

Dividing €2.9 million by 200'000 MWh gives you €15 per MWh (1.5¢/kWh, 55¢ per gallon of gasoline, US$4.30/GJ).

This is pretty far below where current European and US solar PPAs are coming in, and about half of what Lazard gives as the low end. So, either I'm missing something fundamental, or one or more of my assumptions above is inapplicable to most of Europe and the US. Here are my best guesses about what it is:

- A lot of European plants are being built in very polar and cloudy countries like the Netherlands and Germany, so the capacity factor is much worse than 25%. Like 10%.

- The US has massive import tariffs which more than double the price of solar panels.

- Maybe balance-of-plant costs have fallen far behind the precipitous drop in solar module prices, so maybe currently the cost of modules is only 15% of the plant cost instead of the usual 30% or so.

- Maybe lenders consider solar energy projects risky and so demand higher interest rates than the 5% or so that is usual for utility bond flotation.

What do others think?

[ZeroGravitas]

The graph on page 17 tracks the historical costs and shows a massive blip for the top of the price range over the last two years, even as the low end continues the trend of dropping lower.

It's really only that last couple of years top end that lets the comparison to gas prices stand. Without it, the document clearly shows that building new solar is as cheap as buying the gas to burn in an existing gas plant.

I think Lazard cost estimated have always been US specific, though I didn't see that spelled out explicitly at a glance.