Hong Xu9 min read

Where China's solar chain still makes money: inverters, film and sales abroad

Sungrow's inverters made almost 15 times the gross profit of the four largest module makers in H1 2026. Filings show where China's solar chain still earns.

China made at least 86% of the world's polysilicon, wafers, cells and modules in 2024 (IEA PVPS Task 1, October 2025, pp. 7 to 11). Its makers' first-half 2026 reports show that little of that chain made money. Adding up gross profit stage by stage shows where it went: to inverter makers, to the leading film supplier, and to sales abroad.

Gross profit is revenue less the cost of making what was sold, before selling, research, administration and finance costs. It is not profit, but it shows which stages can charge more than their costs. Figures marked "about" are our calculations from the revenue and cost tables in each company's 2026 interim report; the rest are the companies' own. Our solar PV industry map charts the same margins.

Gross profit along the chain

Silicon: losses at every step

  • Polysilicon. Daqo, which makes only polysilicon, had revenue of ¥622.8m against cost of sales of ¥1.09bn, a gross loss of about ¥464m (2026 interim report, p. 25). Its average price was ¥30.63 a kilogram, below its cash cost of ¥34.75 and about half its full unit cost of ¥60.69 (p. 10).
  • Wafers. TCL Zhonghuan's wafers lost about ¥1.79bn on revenue of ¥5.05bn, a gross margin of -35.54% (2026 interim report, p. 16).
  • Across the silicon chain. Tongwei, which makes polysilicon, cells and modules, lost about ¥2.41bn at gross level on ¥19.65bn of solar products (2026 interim report, p. 144).

Materials and cells: thin margins, one exception

Modules: the most revenue, almost no margin

Together the four made about ¥357m of gross profit on about ¥80.9bn of revenue, a margin of about 0.4%. JA's figure covers modules only; the others cover all their solar products.

Inverters: the best-paid stage

Side by side

Our calculations from the same tables:

  • Sungrow's inverter segment made almost 15 times the gross profit of the four largest module makers combined, on less revenue than any one of them. Against Jinko alone, the most profitable of the four, it made about eight times the gross profit on about half the revenue.
  • First Applied's film made nearly three times as much as those four. Three of them, Jinko, LONGi and Trina, are among its named customers (First Applied p. 16).
  • Aiko's back-contact (ABC) modules made about ¥665m on ¥7.28bn, more than the four combined on less than a tenth of their revenue (2026 interim report, p. 138).

Why some stages kept their margins

Few suppliers

The two best-paid stages each have one or two dominant suppliers. First Applied says its share of the world market for encapsulant film has been stable at about half for years, more than the next three makers combined (First Applied p. 16). Huawei and Sungrow together held more than 55% of the world inverter market in 2024, according to Wood Mackenzie data cited by the IEA's photovoltaic programme (IEA PVPS p. 18).

The silicon stages are oversupplied. By the count of the China Photovoltaic Industry Association (CPIA), China had more than 1,100 GW of capacity at each of polysilicon, wafers, cells and modules at the end of 2025, far above global installations (JA Solar p. 30). Polysilicon makers ran at 33.8% of capacity in the first half (Daqo p. 10).

Spare capacity alone does not explain the difference. Inverter capacity also exceeded 1 TW a year in early 2025, against 589 GW shipped in 2024 (IEA PVPS p. 18).

Products bought on more than price

Buyers judge inverters and film on reliability as well as price. Sungrow says it again topped BloombergNEF's 2026 inverter bankability ranking, at 100% (Sungrow p. 12). HIUV notes that because modules must last more than 25 years, film goes through customer tests, third-party testing and long-term reliability checks before it is used, and some project tenders set requirements for the film supplier (HIUV p. 12). First Applied sells specialised films for back-contact and heterojunction modules (pp. 14, 15) and says it keeps an industry-leading margin even in periods of overcapacity (p. 16).

The module stage shows the same pattern. Aiko's ABC modules earned about 9.1% while JA's mainstream modules earned -1.74%. Technology alone is not enough, though: back-contact modules were more than 65% of LONGi's module shipments (LONGi p. 12), and its solar products still sold at a small gross loss.

Our view: overcapacity explains why the silicon stages lose money, but not why inverters and film do not. The difference is what the buyer pays for. Polysilicon, wafers and mainstream modules are sold by the kilogram or the watt, one maker's much like another's; an inverter or a film is tested for reliability, can be specified in tenders, and comes mostly from one or two leaders.

Abroad paid better at every stage that reports it

Every company on our solar map that splits its margins by region did better abroad than at home in the first half, with one company left out for the reason given below:

  • Wafers. TCL Zhonghuan, whole company including its chip wafers, lost less on exports: -11.52% in China against -3.97% abroad (TCL Zhonghuan p. 17).
  • Glass. Flat Glass, whole company: about -1.4% in China, against about 26.7% elsewhere in Asia and about 24.3% in North America (Flat Glass p. 119).
  • Film. First Applied, main business: about 13.7% in China, 23.0% abroad (First Applied p. 140).
  • Cells. Drinda: 4.21% in China, 7.89% abroad (Drinda p. 14).
  • Modules. Aiko, main business: about 1.3% in China, 11.9% abroad (Aiko pp. 138, 139). Jinko, whole company: about -3.3% in China, 5.6% abroad (Jinko p. 165). JA Solar, main business: -2.86% in China, 3.02% abroad (JA p. 16). LONGi, solar products: about -0.5% in China, 1.1% abroad (LONGi p. 176). Trina, whole company including storage and services: about 7.0% in China, 8.1% abroad (Trina p. 206).
  • Inverters. Kehua Data, main business, about half of it solar and storage and two fifths data centres: 19.89% in China, 36.37% abroad (2026 interim report, p. 18).

These are whole-company or segment figures, not single products, so part of each gap may come from a different product mix at home and abroad. Daqo, Sungrow and Sineng do not split margins by region, though Sungrow took 73.40% of its revenue from outside mainland China and Sineng 60.52% (Sungrow p. 35; Sineng p. 22). HIUV's overseas sales, at ¥17m, were too small to compare (HIUV p. 159). Tongwei is left out because its split mixes products: its solar sales include polysilicon, whose buyers, the wafer makers, are almost all in China (IEA PVPS p. 8), so the gap between about -18.7% at home and 3.4% abroad says more about what it sells where than about prices (our calculation from Tongwei p. 144).

The companies give three reasons:

  • Demand at home fell. China installed 72.07 GW of solar in the first half, down 66%, after the move to market-based power prices made project returns less certain and against a high base a year earlier (LONGi p. 10).
  • Prices abroad held. Xinyi Solar says overseas solar glass prices kept a premium, helped by rising international energy and freight costs, and that its plants outside China help it avoid tariffs and earn higher margins (Xinyi p. 42).
  • Makers chose where to sell. Jinko says it cut shipments to low-price, loss-making markets and sent about 70% of its shipments to overseas markets with higher margins (Jinko p. 10).

The premium has a price. China ended the export VAT rebate on solar products on 1 April 2026; exports surged beforehand, and in May module export volumes fell 26.7% from April (Jinko p. 9). The US has extended its duties to Chinese makers' plants in Southeast Asia (Jinko p. 8), with preliminary duties in 2026 on cells and modules from India, Indonesia and Laos (JA p. 30), and Jinko cites rules favouring local manufacturing in India and Europe (Jinko p. 8).

What could move the money

  • A floor under polysilicon. On 6 August eight polysilicon makers, Daqo among them, pledged not to sell below full cost as calculated under a new industry standard (Daqo p. 10). Daqo's full unit cost in the half, ¥60.69 a kilogram, was about twice its selling price.
  • Efficiency rules. From 1 January 2027 mandatory national standards set minimum module efficiencies of 23.2% for TOPCon and heterojunction designs and 23.5% for back-contact (Trina p. 13).
  • Moving into the stages that pay. TCL Zhonghuan is converting 20 GW of cell and 25 GW of module capacity to back-contact after taking control of Yidao New Energy (TCL Zhonghuan p. 12). Daqo plans a ¥6bn base for storage systems, solid-state transformers and other power equipment for AI data centres (Daqo p. 9). Trina's storage business earned about 21.0% on ¥2.47bn, against about 1.3% for its solar products (Trina p. 206).

What it means

Our view: in China's solar chain, scale has not bought margin. The stages China built largest, polysilicon, wafers and mainstream modules, sold at or below cost in the first half of 2026, while a concentrated film maker and the two inverter makers kept healthy margins. A price floor upstream holds only if wafer and module prices rise with it, because wafer makers at a 35% gross loss and module makers at around zero cannot absorb a higher input cost. The customers who pay more are abroad, and that is where the barriers are rising fastest.

Follow the chain on our solar PV industry map, from polysilicon and solar wafers to solar cells, solar glass, encapsulant films, solar modules and solar inverters.

Continue reading