The U.S. sets a floor price for polysilicon, and the tech competition begins to rewrite the cost curve

On August 6, Trump signed a Section 232 announcement targeting polysilicon and its derivatives. The news headline can easily be summarized as "The U.S. imposes an additional 15% tariff on photovoltaic products," but focusing solely on this tax rate underestimates the strength of this policy.

The U.S. has simultaneously provided four clear minimum import prices: polysilicon at $21/kg, polysilicon ingots and wafers at $100/kg, solar cells at $0.22/W, and modules at $0.38/W. Related measures will take effect from December 4, and downstream polysilicon derivatives will also incur an additional 15% ad valorem tax; at the same time, the Department of Commerce has been authorized to exchange tariff exemptions for companies that build polysilicon, ingot, wafer, and cell production capacity in the U.S.

The implications of this set of measures are clearer than traditional tariffs: the U.S. is artificially setting a price boundary that can support domestic manufacturing, and then using investment exemptions to pull companies into this supply chain.

Putting the floor price and the spot market together makes the policy intent more intuitive. On August 5, Bernreuter Research reported that the global average spot price for polysilicon was approximately $4.78/kg, while the price for Chinese n-type material, excluding VAT, was about $4.13/kg; the average price for non-Chinese polysilicon, according to InfoLink, was around $18.5/kg. The $21 floor price set by the U.S. is even higher than the current average quotes for non-Chinese materials.

Beyond tariffs, the U.S. is beginning to directly reshape the pricing system.

The biggest difference between the minimum import price and ordinary tariffs is that it directly establishes a transaction lower limit in the market. According to the White House announcement, importers must prove that the selling price of the relevant products in the U.S. during their first independent transaction is not lower than the minimum import price; if the declared import price is below the floor price, they must make up the difference with specific tariffs. Importers and their affiliates that provide false materials or seriously violate certification requirements may even be permanently banned from importing the relevant products.

Therefore, $0.38/W is not a symbolic reference line. It will become the compliance price boundary that must be faced when entering the U.S. market, and the U.S. Department of Commerce can continue to adjust this set of floor prices based on market conditions.

At the same time, InfoLink reported that the average price for non-Chinese market TOPCon modules was about $0.115/W, while Southeast Asian modules aimed at U.S. projects were around $0.27/W, and the DDP average price for U.S. assembled TOPCon was about $0.31/W. The three sets of quoted figures are not entirely consistent and cannot be directly subtracted as landed costs, but the magnitude is sufficient to illustrate the issue: the U.S. aims to protect not just a few percentage points of tax difference, but a sufficiently large manufacturing profit margin.

Figure 1｜Polysilicon: U.S. Minimum Import Price vs. Spot Price Reference

Note: Price standards are explained in the figure; spot data is from August 5, 2026, and the policy floor price is the announced value on August 6.

Figure 2 | Components: Minimum import price in the U.S. and reference quotes from different markets

Note: FOB, DDP, and general market quote standards differ and are only used to observe price levels, not for direct calculation of corporate net profits.

Why polysilicon: 2.4% semiconductor demand tied to 97.6% photovoltaic demand

Polysilicon appears in both the photovoltaic and semiconductor supply chains, but the two materials cannot be simply equated. The Semiconductor Industry Association (SIA) provided a crucial set of data in its 232 investigation opinion: global photovoltaic-grade polysilicon demand is expected to reach 1.3794 million tons in 2025, while semiconductor-grade is only 33,500 tons, accounting for just 2.4% of total demand.

The purity, cost, and customer systems of the two are also different. SIA disclosed that semiconductor-grade polysilicon typically requires a purity of 11N or higher, while photovoltaic-grade generally ranges from 6N to 10N; industry estimates suggest that the production cost of semiconductor-grade materials can be up to 30 times that of photovoltaic-grade. In other words, the actual amount of polysilicon consumed by semiconductors is small, but it requires higher purity and more complex process control.

However, these two markets are interconnected in terms of factory economics. The output of high-purity materials is small, and if a company relies solely on semiconductor-grade products, it is difficult to achieve sufficiently high capacity utilization and economies of scale. The White House announcement also clearly articulated this relationship: the large-scale demand for solar-grade polysilicon can help U.S. manufacturers maintain the unit cost required for producing semiconductor-grade materials.

Photovoltaics provide scale, while semiconductors provide strategic value. This is also why this policy did not stop at chip equipment or wafer stages, but continued to push competition down to the basic material level.

The U.S. is fastest in supplementing components, but the most challenging supplements remain upstream

Over the past four years, U.S. photovoltaic manufacturing has indeed been expanding. The U.S. Department of Energy previously estimated that domestic component annual production capacity could approach 40GW by 2026; as of the first half of 2024, the manufacturing capacity newly added or announced across the entire industry chain after the IRA exceeded 95GW, with new component capacity nearing 42GW.

However, rapid expansion of components does not mean the entire chain has been rebuilt. The data released by the White House better illustrates the shortcomings: the U.S. share of global polysilicon production capacity has dropped from about 50% in 2005 to less than 2% in 2024; silicon ingots, wafers, and solar cells still "almost entirely rely on imports." Reuters also pointed out that currently, the U.S. has only two polysilicon plants, belonging to Hemlock Semiconductor and Wacker Chemie China's concentration is at the other end. The IEA's "Energy Technology Perspectives 2026" shows that in 2024, China will account for about 90% of the global upstream capacity for photovoltaic materials such as polysilicon and silicon wafers, with a component production share of about 80%. Even under the IEA's policy scenario, by 2030, China's share in upstream polysilicon and silicon wafers will still exceed 80%, and components will still exceed 70%; the United States may come close to self-sufficiency in components, but upstream reconstruction is clearly slower.

This explains why the U.S. is pushing its policy forward from components. The assembly side can quickly increase nominal capacity within a few years, while the materials, wafers, and battery segments require longer construction cycles, process accumulation, and supplier support.

China's cost advantage must also account for excess capacity.

If we only attribute the low prices of Chinese photovoltaics to subsidies, it still cannot explain today's market. The IEA's earlier supply chain research has clarified the cost differences: in 2020, the cost of manufacturing crystalline silicon components in the U.S. was 30%-40% higher than in China, with labor accounting for 22% of manufacturing costs in the U.S. and 8% in China; due to gaps in the industrial chain, the U.S. also needs to bear about 11% in additional import-related costs. Although the data is from an earlier year, the mechanisms it reveals have not disappeared—scale, industrial clusters, energy, and supporting efficiency jointly determine the cost curve.

At the same time, the current low prices in the Chinese market have another factor: excess. On August 5, the price of China's n-type polysilicon was about 31.5 yuan/kg, which Bernreuter converted to an approximate price of $4.13/kg excluding VAT, clearly indicating that the price is already below cash costs. The top ten global photovoltaic manufacturers, as reported by the IEA, are expected to collectively lose about $4.5 billion in 2024.

Therefore, the price differences seen today do not equate to a fully replicable "normal cost." They include both the efficiency advantages that China has long established and the loss-driven competition following industry excess. The U.S. sets a minimum import price, essentially temporarily isolating domestic companies from this global low-price curve.

However, price protection can only create profit space initially. Capacity utilization, yield rates, equipment maintenance, engineer experience, and upstream-downstream collaboration still rely on real production to form. Whether the U.S. can convert "protected prices" into "competitive capacity" will take at least several investment cycles to verify.

Global photovoltaics may enter two or even multiple cost systems.

This institutional design has a detail that is easily overlooked: it does not require all capacity to return to the U.S. The White House allows for tariff incentives for companies that commit to building related factories in the U.S. through an "onshoring plan," and has set different arrangements for some trading partners. If other countries adopt minimum import prices that are roughly equivalent to those of the U.S., the U.S. can also adjust the application of MIP and tariffs for these partners.

A more likely structure is that U.S. domestic manufacturing and allied supply chains jointly form a protected high-price market; outside the U.S., a price system closer to global supply and demand will continue to operate. The supply chain will thus shift from "global lowest cost priority" to "regional security + traceability + policy compliance." For Chinese enterprises, the impact is difficult to simply summarize as a decline in exports. The U.S. market has already been burdened with restrictions such as anti-dumping, countervailing duties, and the UFLPA, making direct imports of photovoltaic products from China no longer the main channel. What needs more attention are two things: first, whether non-U.S. markets can absorb the excess capacity that has been redirected; second, whether the overseas production capacity of Chinese enterprises in Southeast Asia, the Middle East, and other regions can meet the increasingly complex requirements for origin and supply chain traceability.

If the high-priced U.S. market is effectively isolated, global excess capacity may be more directed towards Europe, the Middle East, and Latin America, thereby increasing price pressure in these regions. At the same time, companies with non-Chinese sources of polysilicon, local processing capabilities in the U.S., or compliant supply chains may gain a temporary price premium created by policy.

What is worth monitoring next is not the tariff slogans.

Whether this policy can ultimately rebuild the U.S. upstream supply chain, I would prefer to look at several sets of verifiable data: the actual commencement and production of new polysilicon, silicon ingot, wafer, and battery projects in the U.S.; whether capacity utilization and yield can continue to improve; whether the transaction prices of U.S. components maintain a long-term difference from global market prices; whether higher equipment costs slow down U.S. photovoltaic installations; and the speed of exit, overseas migration, and market diversion of China's excess capacity.

A price bottom can be established in an announcement, but manufacturing capacity cannot be generated in an announcement. The weight of this U.S. policy will ultimately be reflected in these data