It's amazing how they have been wrong on this and still continue making the same assumptions every single year (see other comment).

Either they are wrong or they are indirectly predicting the bancruptcy of the PV industry at large. I'd bet at the former, all that additional production capacity will be used one way or another.

Yeah, to be fair solar PV can be hard to estimate in countries with lots of feed in solar panels because it is decentralised.

Module production is quite centralised, with about 80% of it in China, and another 10% or more in South-East Asian countries, set up there by Chinese companies to get around the US tariff imposed in 2019 or so. (Korea produces some of the cells but doesn't assemble modules, AFAIK. Japan, also a few cells.)

The Europeans want to re-animate their domestic manufacturing for energy security reasons, after letting it die five or six years ago. First Solar is trucking on in the USA. India has ambitions to expand its production a lot. Not too many other places, to my knowledge.

Module production may ultimately become decentralized, for the same reason production of window glass is decentralized: transportation costs.

> Annual PV capacity production of around 400 GW in 2023

How does that make sense. 400GW Peak? Average?

Or are they talking GWh

If the IEA can't use the right units in its reports, what hope is there.

Nameplate capacity. E.g. if panels are all 400W panels, it’s 1 billion panels.

The IEA is using the correct unit. They are talking about "annual solar PV capacity additions" which is measured in Watts.

It's correct but not very useful, as it doesn't inform about the real energy output (and who cares about the nominal/theoretical output). PV in a desert in US Southwest produces vastly more energy than say PV in Germany.

The difference is around a factor of two between the US Southwest deserts and the German north sea coast [1]. So tracking installed peak capacity is not an ideal metric, but it is very simple and gets the order of magnitude right.

[1] https://globalsolaratlas.info

It's not supposed to measure real energy output, it's about predicting the capacity increase. What the actual energy output will be is a lot harder to predict and there is a lot more uncertainty regarding those numbers. Regular power plants are denominated in the exact same way, even though the actual energy output will be lower.

Real energy output is just one of many possible ways of measuring PV, but it’s even less predictable than number of panels. Fixed installations produce less power per year than 1 or 2 axis tracking etc. Different locations mean different amounts of transmission losses etc.

But even the best model is going to get blindsided by politics. Europe would have been well severed by massively investing over the summer in PV to reduce it’s dependence on Russian natural gas. They didn’t do so but I was very much expecting a push for the fastest to install alternative to natural gas turbines which is PV.

Just want to add that I may be underestimating 2030 production if perovskites take off.

Roll-to-roll printing on flexible plastic film is a lot cheaper than refining silicon, slicing it up, and embedding it in glass to make PV modules.

The unknowns are all upside risk with PV and wind. Not so much with fossil fuels.

I think they are being conservative with their estimates and being pessimistic on announcements.

Perhaps they are accounting for the China slowdown? PV has flattened in China.