Excluding the effects of impairment, the performance slightly exceeded expectations: the company achieved net profit of 4.35 billion yuan (RMB 4.35 billion; of which the polysilicon sector was 3.66 billion yuan and the power plant sector) last year, down 67% year-on-year, in line with the performance forecast (42-4.4 billion yuan). Among them, the loss for the fourth quarter was 730 million yuan, mainly due to inventory impairment of 286 million yuan and depreciation of fixed assets of 520 million yuan. After excluding the effects of impairment, the profit was 120 million yuan, higher than our forecast of 0.4 billion yuan. Of these, the power plant sector made a profit of 260 million yuan. As costs were still high, the polysilicon sector lost a small amount. The company's polysilicon production/sales volume last year was 19.1/203,000 tons, an increase of 52%/90% year-on-year. It plans to produce 28-300,000 tons this year, and there are no plans to increase polysilicon production capacity. At the end of last year, the cumulative installed capacity of the power plant increased by 0.6 GW to 3.2 GW over the same period last year, and it is planned to increase to 4-5 GW by the end of this year. The construction scale of the power plant that confirmed revenue last year was 2.2 GW, and the plan is 2-3 GW this year; inverter shipments doubled to 15 GW over the same period last year, and the plan is to increase to 30 GW this year. Since the A-share IPO is still ongoing, there will still be no dividend payout in 2023, and we expect to resume dividends after the IPO is completed.

After full production, the cost of polysilicon has dropped drastically, and the proportion of N-type materials has increased rapidly: as production capacity climbed, the company's polysilicon production cost fell to 54,000 yuan/ton in the fourth quarter, down from 70,000 yuan/ton in January, and fell to 48,000 yuan/ton in February with full production at all three bases. The company expects that there will still be room for a decline of 30,000 yuan/ton in the future. The company's share of N-type materials has now reached 65-70%, and plans to increase it to 85% this year. As the share of downstream N-type materials increases, the N/P price spread will continue to widen, and the competitive advantage of traditional leaders, which have a clear advantage in the share of N-type materials, will become more obvious.

Demand exceeds expectations and is expected to raise the bottom of polysilicon prices: in January-January, exports of new PV installations/battery modules reached 37/49 GW, a sharp increase of 80%/54% year on year. Global module production schedule in March was about 55 GW, a sharp increase of 50% over the previous month, all exceeding market expectations and driving the recent price stability of PV products. We expect the global growth rate of new installed capacity to be close to 30% this year, clearly exceeding our previous expectations of 20%. Polysilicon prices have stabilized temporarily after a slight rebound since this year, but new production capacity is still being released, so we expect prices to fall again, but demand exceeds expectations and is expected to raise the bottom of prices. We expect the company's average tax-inclusive sales price of polysilicon to stabilize at around 60,000 yuan/ton in 2024-26 (108,000 yuan last year).

It is expected that the company's polysilicon sales price will be at the leading level in the industry, driven by a high proportion of N-type materials, and the cost is close to Dazuan Energy and there is still room for decline after a sharp drop, and it will also be highly competitive, so the polysilicon business will still be profitable at the bottom of the cycle. We expect a net profit of 18.8/22.5/2.79 billion yuan in 2024-26, of which the steadily growing power plant sector will contribute 9.3/12.6/1.55 billion yuan. The target price was lowered to HK$12.80 (previously HK$13.50) based on a price-earnings ratio of 9 times 2024 (compared to the 13x discount we gave to GCL Technology, which is lower in cost, to reflect higher performance uncertainty).