Your browser is not supported. Please update it.

11 January 2023

Chinese solar manufacturing hurtles toward overcapacity

Last year around 330 GW of photovoltaic modules were produced globally. The global markets can absorb and install that much this year – so far polysilicon, cell and wafer cost falls of 50%, 25% and 25% have been met with a module cost fall of only 5% to 10%, since demand from installers is very high. This week’s EnergyTrend price tracking shows another 3% module decline and 20+% polysilicon cost decline.

We’ve predicted a general fall in photovoltaic prices including module prices. One interesting alternative is that we’d move from polysilicon makers charging well above their production cost (actually, seven times their production cost) in 2022, to the whole supply chain charging above production cost, because consumers are so eager and mostly sustained the high prices of 2022 which were inflicted by the polysilicon makers.

This seems unfeasible in the scale of further production expansion in 2023 which will take output well beyond 330 GW, to maybe 500 GW. Polysilicon will be around 1.6 million tons of output, given that it’s presently at 100,000 tons per month.

We’ve predicted that polysilicon will fall to $18.3 per kilogram by the end of the year, but Silicon Industry Branch reports that this week Zhonghuan’s polysilicon price already fell to $17.7, which is 10% lower than the price report by EnergyTrend. Compared to a price of $40 only a couple of months ago, going by a rough figure of 2.7 grams per Watt, this means the production cost of 1 kW of photovoltaics has fallen by $60 in a matter of weeks.

Silicon Industry Branch also implies that the cost will fall to $9 per kilogram this year, and that module prices will fall to between $192 and $207 per kW, with vicious cost competition. Already POWERCHINA’s 2023 module procurement bidding varies from $234 per kW to $277 per kW.

Incidentally, the Silicon Industry branch is now reporting polysilicon prices once more, having ceased its four-month pause on that report, which was seemingly part of a government effort to limit the price gouging or price speculation going on with polysilicon.

A full monthly picture of Chinese photovoltaic production output for 2022 isn’t yet available, but not only is polysilicon now produced to the tune of 100,000 tons a month, cell and wafer production was at around 35 GW to 36 GW in November, which annualizes to 432 GW.

If anything 500 GW to be made in 2023 is too conservative, the product of unwillingness (on the part of this writer as much as anyone else!) to believe in what the actual numbers are showing. It strongly implies that 2024 will see over 400 GW installed, or alternatively a host of bankruptcies among manufacturers, or both. If the manufacturers keep up the 2022 pace of expansion, or even half of that pace, surely they will cause overcapacity and a supply glut in 2024. Eventually some part of the demand side will fail to keep up. If it’s not financing, in the context of global fuel trade disruption and high energy costs, then it will be workforce – for a year or so while installers hire and train – or transmission line constraints, which would take longer to catch up, or the supply of lithium for supporting battery energy storage.

One caveat is that existing wafer and cell production capacity can be decommissioned as obsolete under such pressures. PERC is slightly less competitive than its immediate successors, heterojunction and TOPCon. Something like 600 GW of cell production capacity was announced or built in 2022, but much of this was replacement.

Another caveat is that 2024 demand may falter at 400 GW. If the past few paragraphs were written around 36 GW of production November, then we have to mention 27.1 GW manufactured in the wafer segment in December, a 25% drop. However, much of this is due to China’s heavy Q4 and December skew for installations meeting end-of-year deadlines. That pushes up the November figure, when it’s made just in time, and depresses the December figure, when it’s too late. Module output was also down 18.8%, to “only” 26.8 GW, but there’s a lot of module production outside of China which there isn’t for wafers.

So if you think annualizing 36 GW to 432 GW produced figures that were too high – maybe let’s annualize the average of 36 GW and 27 GW, which is 31.5 GW, to reach 378 GW produced in 2023 – bare minimum – implying at least 400 GW, not at least 500 GW, installed in 2024. Either way, the growth year-on-year is still at the very least 30%, just as the growth in global installations from 2021 to 2022 was 36.5% or so.

Of the millions of tons of announcements of new polysilicon capacity, around two-thirds come from companies with no prior polysilicon facilities. Assuming half of the announcement are cancelled and all cancellations come from this camp, that still means one-third of new capacity will be entrants. In last week’s issue we mentioned that 35 Internet and Allwin Telecommunications, coming from software and mobile phones respectively, will enter solar cell manufacturing. While such cross-border efforts are common in China, it’s remarkable considering that solar manufacturing typically has low margins, due to the necessity of constant reinvestment in updated production lines. This week there’s yet another new entrant, engineering services company Suzhou Shijing, with a $1.65 billion investment in 24 GW of TOPCon cell production, of which 18 GW is to be built from April to November.

The biggest share of 2023 manufacturing expansion will be TOPCon, with its high compatibility with PERC production lines. For this type, PV Infolink predicts 350 GW of production capacity by the end of the year and 98 GW output during the year.

Heterojunction manufacturing stocks are up 8.78% in the past month in China. The technology boasts the best efficiency potential in silicon PV thanks to high bifaciality, but now the technology’s current high production costs are being beaten down by research and instead potential production advantage, owed to the technology’s ability to use especially thin wafers, are starting to be realised. This year will see at least 10 GW produced. Recently the first heterojunction modules were deployed in a small project.

Heterojunction’s thinness – recently one company said it’s planning 100μm wafers, compared to the current industry standard of 155μm – also makes it the best candidate for silicon-perovskite tandems. There are several Chinese companies pursuing perovskites, ranging from some with a respectable research base and years of involvement, such as Utmolight and Microquanta, to others that are recent entrants with $10 million of freshly raised finance. Mostly, these appear to be pursuing perovskite-only thin film for the BIPV segment.

Utmolight claims to be running a pilot line right now with mass production next year, and similar statements are being made by a few other companies, but such claims have been made by perovskite companies for several years’ worth of “next year”. At least their claims or easily halving cost per Watt, if not better, are credible, based on the necessary raw materials and processes involved.