As the spot market price for solar-grade polysilicon remains stuck at $34 per kilogram, near December’s 10-year high, solar manufacturers have been making more and more long-term orders.
All of these deals are based on monthly price negotiations, and their value is usually reported as if they were based on the current average selling price – while we expect they will get a better deal than on the spot market, the polysilicon suppliers have the upper hand and primary aim of these agreements is for manufacturers to secure a guaranteed supply of raw material. Prices for solar wafers are almost 90% higher than in January 2021, with modules up 20% in twelve months – and rising again this month by up to 1% per week.
Long-term polysilicon supply contracts are by no means a new thing, but they had been declining in late 2019 and early 2020. Since late 2020 they have been increasing in length, scale and frequency. Where a few years ago a typical contract would be a few tens of thousands of tons, lasting two years, now they have a duration four or five years and often surpass 100,000 tons each.
A random selection of just a few deals gives us an idea of the situation:
- TBEA subsidiary Xinte Energy’s order from LONGi for 270,000 tons from January 2021 to December 2025, signed in December 2020.
- JA Solar and GCL’s 145,800-ton contract signed in May 2021, running through to June 2026.
- Daqo New Energy and JA Solar’s contract for about 78,200 ton signed in May 2021, with an end date of December 2025. Not just the price but also the scale of the polysilicon purchases were stated to be subject to monthly negotiations.
- WACKER’s order from JinkoSolar for 70,000 tons, in an agreement signed in August 2021.
- Xinte Energy’s order from Shuangliang for 82,200 tons, from January 2022 to December 2026, with monthly price negotiation specified to be based on market prices from the 25th to 30th of the previous month. This deal was struck in September 2021.
Just this week, there is another agreement – 203,600 tons from Tongwei to LONGi, lasting through to December 2023. However, most deals have been signed to last through to 2026.
According to data collated by Chinese news site Polaris Solar Network, total long-term polysilicon supply contracts have risen to a total of 1.4 million tons. We estimate that this year’s world polysilicon supply will be 721,000 tons, so these deals amount to two years of global output, though with a typical timeframe of five years. Already in August TBEA had announced that its whole output was accounted for throughout the next three to five years.
In our research paper on the polysilicon industry’s expansion, published a few months ago, we tracked expansion announcements equalling 3.3 million tons from twelve leading Chinese companies, with the expectation that these announcements would reach 4 million tons to be built by 2025, but that only 2.3 million tons would be built as the deluge of supply removed the incentive to build more.
Since the report’s publication even more manufacturing announcements have indeed rolled in, including a 200,000-ton factory in Indonesia, with eighteen companies in total announcing expansions, fourteen of them with plans exceeding 100,000.
The sheer scale of polysilicon manufacturing means we are sure to see another polysilicon glut and a collapse in prices.
Compared to 3 months for module factories and 9 months for wafer and cells, polysilicon factories take 18 months minimum to build – none of that is counting subsequent ramp up time, which is three months for the rest and at least 6 months for polysilicon. So that’s two years’ delay from investment to output for polysilicon, but the first expansion announcements were already being made by Asia Silicon, GCL-Poly and Daqo New Energy in Q3 of 2020.
That is the main reason we expect the polysilicon price to clearly decline in mid-2023 – then in just two or three years’ time we will have yet another supply chain mismatch, but with too much polysilicon production capacity instead of too little. If there is 2.3 million tons of production capacity online and fully ramped up in 2026, working at 60% capacity utilization, that’s enough for over 450 GW of silicon PV, clearly far in excess of the 350 GW we expect demand to have reached by then. And that’s just after factoring a 60% capacity utilization, which is moderately low, and also that many announced projects will not end up getting built.