Silicon

China's Silicon Chokehold: The Xinjiang Forced-Labor Shock

The USGS Mineral Commodity Summaries 2026 reports that China accounted for almost 80% of total global estimated production of silicon materials in 2025, with Chinese silicon-metal output of roughly 4.0 million tonnes against a world total near 4.6 million tonnes. On the polysilicon side — the ultra-purified feedstock for both solar cells and semiconductors — Bernreuter Research found China had expanded capacity to 3.25 million tonnes by the end of 2024, about 93.5% of global output, with nine of the world's ten largest polysilicon producers based in China (pv magazine / Bernreuter Research, Nov 2025).

On June 21, 2022, the rebuttable presumption under the Uyghur Forced Labor Prevention Act (UFLPA) went into effect — 180 days after the law's December 23, 2021 enactment — requiring U.S. Customs and Border Protection to presume that any goods made wholly or in part in the Xinjiang Uyghur Autonomous Region (XUAR), or by a listed entity, are produced with forced labor and to block their entry unless importers provide clear and convincing evidence otherwise (Skadden, June 2022). Polysilicon and other silica-based products were named a high-priority enforcement sector from the outset, alongside cotton, apparel, and tomatoes.

The Department of Homeland Security's Forced Labor Enforcement Task Force has since designated dozens of Xinjiang-linked silicon and solar entities to the UFLPA Entity List, including a January 14, 2025 batch of 37 new entities — the largest single addition to date — naming Xinjiang polysilicon and metallurgical-silicon producers such as Shuangliang Silicon Materials (Baotou), Baotou Meike Silicon Energy, and Xinjiang Energy Group (Miller & Chevalier, Jan 2025). The Responsible Sourcing Tool's 2025 Polysilicon Commodity Report confirms polysilicon is documented as produced with forced labor in Xinjiang under state-sponsored labor-transfer programs, citing the State Department's 2025 Trafficking in Persons Report and the Department of Labor's List of Goods Produced by Child Labor and Forced Labor.

Why it matters: Polysilicon sits at the base of both the solar and semiconductor supply chains. Even as Western module makers built parallel non-Xinjiang supply lines, CSIS's three-year UFLPA review (Aug 2025) found Xinjiang's share of global solar-grade polysilicon capacity fell from ~41% in 2021 to ~24.8% by mid-2025 — but Xinjiang's share of China's metallurgical-grade silicon rose to 53%, meaning the upstream raw-silicon layer remains heavily exposed to forced-labor risk even where refined polysilicon has diversified.

1. Two product streams, one province: metallurgical silicon vs. polysilicon exposure

Silicon's forced-labor exposure runs through two separate but linked product streams. Metallurgical-grade silicon (~98% purity) — the primary smelted product that feeds aluminum alloying, silicone chemistry, and further refining — is dominated by China at nearly 4.0 million of the world's 4.6 million tonnes in 2025 (USGS MCS 2026), and per the CSIS three-year review, Xinjiang alone now accounts for 53% of China's metallurgical-silicon output (CSIS, Aug 2025). Polysilicon — metallurgical silicon further purified to 6N–11N for solar cells and semiconductors — is separately dominated by four Chinese producers (Tongwei, GCL Technology, Daqo New Energy, Xinte Energy), which together held roughly 65% of global 2024 output, with nine of the world's top ten producers based in China (Solar Power World, citing Bernreuter Research, Dec 2025). UFLPA enforcement targets the polysilicon/solar layer far more heavily than the upstream metallurgical-silicon layer, which is why Xinjiang's polysilicon share has fallen even as its metallurgical-silicon share has risen.

2. The UFLPA Entity List mechanics and enforcement escalation

The UFLPA created a standing UFLPA Entity List maintained by DHS's Forced Labor Enforcement Task Force; any importer with goods linked to a listed entity faces the same rebuttable presumption of forced-labor origin as goods made wholly in Xinjiang (State Department UFLPA fact sheet, Jan 2025). The list has grown from 20 entities at inception in mid-2022 to roughly 144 entities by mid-2025 (LandedFees UFLPA tracker, June 2026), with the January 14, 2025 batch of 37 additions — the single largest expansion — specifically naming Xinjiang polysilicon and metallurgical-silicon producers including Shuangliang Silicon Materials, Baotou Meike Silicon Energy, and Xinjiang Energy Group (Miller & Chevalier, Jan 2025). In August 2025, the Forced Labor Enforcement Task Force added five new high-priority enforcement sectors — caustic soda, copper, lithium, red dates, and steel — broadening the UFLPA's reach beyond its original cotton/polysilicon/tomato/apparel focus (USTR press release, Aug 2025).

3. China's polysilicon overcapacity blunts the diversification narrative

Even as Western buyers have diversified module assembly away from Xinjiang, China's overall polysilicon dominance has not eroded — it has intensified. Bernreuter Research found China expanded nameplate polysilicon capacity to 3.25 million tonnes by end-2024, about 93.5% of global output, against global demand of roughly 1.1 million tonnes — a capacity-to-demand ratio approaching 3x (pv magazine / Bernreuter Research, Nov 2025). Beijing's own competition regulator halted a proposed CNY 50 billion (~$7 billion) consolidation fund among Tongwei, GCL, Daqo, Xinte, East Hope, and Asia Silicon in January 2026 intended to acquire and idle roughly one-third of domestic capacity, sending polysilicon futures down 9% in a single trading session (pv magazine, Jan 2026).

Price Collapse: Silicon Metal and Polysilicon Diverge From Crisis Narrative

Per USGS MCS 2026, the January–October 2025 average U.S. price for silicon metal was about 21% below the 2024 annual average, and in April 2025 Chinese silicon-metal export prices hit their lowest level since November 2016. USGS attributes the decline to oversupply, weak demand from the aluminum-alloy and silicone-chemical industries, and the overhang of polysilicon stocks. On the solar-grade polysilicon side, PV Tech (Dec 2025) reports average polysilicon prices collapsed from $32.7/kg in February 2023 to $4.4/kg in May 2024 as Chinese capacity additions outran a slowing solar build-out.

The price picture is starkly bifurcated by grade and geography. IMARC Group's June 2026 tracking shows polysilicon at $4.73/kg in Northeast Asia versus $25.29/kg in North America (IMARC Group, June 2026) — a more than 5x premium reflecting U.S. Section 301 tariffs (50% on Chinese polysilicon, effective January 1, 2025), IRA/CHIPS-linked domestic-content incentives, and the scarcity of non-Chinese semiconductor-grade capacity. Semiconductor-grade polysilicon (11N+ purity) trades at a substantial premium to solar-grade material and is not published on a comparable spot index.

No LME listing: why silicon prices on assessed benchmarks, not exchange futures

Unlike copper, aluminum, or nickel, silicon metal and ferrosilicon are not listed products on the London Metal Exchange. Price discovery instead runs through periodic assessments published by price-reporting agencies: Argus Media's "Silicon 4-4-1 min 99% Si ddp Europe works" assessment is a widely referenced European benchmark for standard commercial-grade silicon metal (99% Si, with maximum 0.4% iron, 0.4% aluminum, and 0.1% calcium — the "4-4-1" designation) (Argus Media, Silicon 4-4-1 assessment). Fastmarkets publishes a parallel "Silicon grade 4-4-1, 99% Si min, in-warehouse Rotterdam, €/tonne" assessment (Fastmarkets, Silicon 4-4-1 Rotterdam), while CRU Group assesses ferrosilicon and silicon metal as part of its bulk ferroalloys and metallics price-reporting services used across the steel, foundry, and aluminum industries (CRU Group, Silicon commodity page). USGS itself sources its official ferrosilicon and silicon-metal price series from CRU Group (weekly transaction-price averages) and S&P Global Platts Metals Week (monthly mean import prices), not from any exchange-traded futures contract (USGS MCS 2026).

Ferrosilicon and silicon metal: diverging 2022–2025 price paths

USGS's own U.S. price series shows a sharp round-trip: average silicon-metal price (cents per pound of contained silicon) spiked to 361.86¢/lb in 2022 amid post-COVID demand and energy-cost shocks in Europe, then fell to 179.69¢/lb in 2023, 170.34¢/lb in 2024, and an estimated 130¢/lb for 2025 — a decline of nearly two-thirds from the 2022 peak. Ferrosilicon (75% grade) followed a similar arc: 312.10¢/lb in 2022 down to 131.96¢/lb in 2024 and an estimated 140¢/lb in 2025 (USGS MCS 2026). USGS discontinued its separate 50%-grade ferrosilicon spot-price series in April 2022 as market liquidity concentrated in the 75% grade.

The US Response: CHIPS Act, Domestic Polysilicon, and a Section 232 Probe

President Biden signed the CHIPS and Science Act into law on August 9, 2022, appropriating $52.7 billion for semiconductor manufacturing incentives ($39 billion), R&D ($11 billion), and a 25% investment tax credit — part of $280 billion in total authorized spending (Encyclopaedia Britannica). On October 21, 2024, the Commerce Department announced preliminary terms for a $325 million CHIPS award to Hemlock Semiconductor (majority-owned by Corning) to build a new hyperpure semiconductor-grade polysilicon facility in Hemlock, Michigan — Hemlock is the only U.S.-based maker of hyperpure semiconductor-grade polysilicon (Commerce Dept, Oct 21 2024; Bloomberg). Construction was slated for 2025–2026 with production beginning in 2028.

Wacker Chemie, the world's largest producer of semiconductor-grade polysilicon, has expanded its Tennessee and Germany capacity: per the Wacker Annual Report 2024, the company is expanding semiconductor-grade polysilicon capacity in Burghausen, Germany, with a new etching line ("Etching Line Next") delivering first output in Q2 2025 to enable chips at the 3nm node and below. Wacker's Charleston, Tennessee site — its largest single investment in the Americas — anchors its North American polysilicon and silicone footprint (Wacker Tennessee overview).

Not every domestic restart succeeded. REC Silicon restarted granular polysilicon production at Moses Lake, Washington in Q4 2023 under a 10-year, ~$3 billion take-or-pay agreement with Hanwha Qcells, but on December 30, 2024 shut the plant down again after failing to meet Qcells' purity specifications, ending U.S. polysilicon output at that site entirely (Solar Power World, Dec 2024; REC Silicon Annual Report 2024). REC Silicon has since pivoted to a pure-play silicon-gas producer for battery applications.

On July 1, 2025, the Commerce Department's Bureau of Industry and Security initiated a Section 232 national security investigation into polysilicon and its derivatives, examining import concentration, Chinese state subsidies, and the feasibility of tariffs or quotas to protect domestic capacity (Federal Register notice, July 16 2025). The investigation must conclude within 270 days (by late March 2026); the Semiconductor Industry Association and clean-energy trade groups both filed comments by the August 6, 2025 deadline (SIA comments, Aug 2025).

Defense & strategic uses — silicon carbide, radiation-hardened chips, and the semiconductor base layer

Silicon's defense criticality operates on two tracks: bulk metallurgical/polysilicon feeding the broader semiconductor and defense-electronics base, and specialized silicon-derived materials — silicon carbide (SiC) and silicon-on-insulator (SOI) — enabling power electronics and radiation-hardened chips that cannot be sourced from adversary supply chains.

1. Silicon carbide (SiC) power electronics in aircraft and radar

SiC's wide bandgap gives it higher breakdown voltage, thermal conductivity, and radiation hardness than silicon, enabling smaller, lighter power converters critical to military aircraft where weight savings translate directly to payload and range. Per aerospace power-electronics analysis, SiC MOSFETs and Schottky diodes operate at higher temperatures and voltages than silicon devices, and flight-tested SiC motor controllers have shown power-conversion losses reduced by up to 50% versus silicon-based solutions. Military radar systems — including active electronically scanned array (AESA) radars — increasingly rely on GaN-on-SiC RF transistors for higher power density and thermal management (EDN). The Air Force Research Laboratory has funded large-diameter SiC substrate development specifically for radar, electronic warfare, and communications applications spanning 300 MHz to 300 GHz (Military Aerospace, 2016).

2. Radiation-hardened silicon-on-insulator (SOI) chips for space and nuclear-survivable systems

On December 21, 2021, the President authorized DPA Title III action "to strengthen and expand the domestic industrial base for radiation-hardened and strategic radiation-hardened microelectronics," citing DoD's need for trusted components that withstand ambient space radiation and proximity to nuclear detonations for satellites and nuclear modernization programs (Dept. of War, Dec 2021). SkyWater Technology's Bloomington, Minnesota foundry uses a 90nm fully-depleted silicon-on-insulator (FDSOI) process — licensed from MIT Lincoln Laboratory — that eliminates the bulk-silicon latch-up current path responsible for radiation-induced failures, as part of a DoD-funded program of up to $170 million under the Trusted and Assured Microelectronics program (SkyWater/MIT Lincoln Laboratory, 2020). On September 6, 2024, DoD awarded $25.8 million to Honeywell (Plymouth, Minnesota) to sustain 90nm strategic radiation-hardened microelectronics fabrication for current and future U.S. space and strategic systems (Dept. of War, Sep 2024). Only three trusted radiation-hardened-by-process foundries exist in the United States (DARPA/MIT Lincoln Laboratory, ERI Summit).

3. Polysilicon as the base layer of the entire defense semiconductor stack

Every silicon-based integrated circuit — from commodity microcontrollers to the most advanced AI and radiation-hardened logic — begins with electronic-grade polysilicon of at least 99.999999999% (11N) purity, per the Semiconductor Industry Association's Section 232 comments. Because DoD weapon systems, satellites, and command-and-control networks depend on semiconductor fabs that in turn depend on this single upstream input, USGS's addition of silicon to the U.S. Final 2025 List of Critical Minerals (published November 7, 2025) formally recognized silicon's cross-cutting national-security relevance (USGS MCS 2026).

4. Historical DPA Title III precedent: SiC MMIC devices for radar

DoD's use of DPA Title III to build domestic SiC device capability predates the current cycle: Fiscal Year 2009 budget documents show DPA Title III funding for "Silicon Carbide (SiC) Monolithic Microwave Integrated Circuits (MMIC) Devices & Materials," intended to enable a new generation of high-frequency, high-power radar, high-temperature sensors, and power-management devices (DoD FY2009 DPA budget justification). This decades-long investment underpins current SiC power electronics in modern electronic warfare and radar systems, including active radar and EW suites on fifth-generation fighter platforms.

Trade flows — UFLPA enforcement redirects solar supply chains, not China's underlying dominance

Unlike antimony or gallium, silicon has not faced a formal Chinese export ban. Instead, the trade story is the United States restricting imports via forced-labor enforcement while China continues to expand output for the rest of the world — producing a supply chain that looks diversified on paper but remains deeply China-dependent in raw metallurgical silicon.

Silicon metal: US import sources, 2021–2024

Per USGS MCS 2026, U.S. net import reliance for silicon metal crossed above 50% of apparent consumption in 2025 for the first time in the five-year data series (it was under 50% every year from 2021–2024). Total U.S. silicon metal imports in 2025 were estimated about 50% higher than in 2024, which USGS attributes partly to importer stockpiling ahead of anticipated tariff and export-restriction actions. Direct U.S. imports of silicon metal from China are minimal — the exposure runs instead through downstream polysilicon, wafers, and finished solar modules manufactured with Chinese or Xinjiang-linked material and assembled in third countries.

UFLPA enforcement: solar/polysilicon dominates detention value

Reuters' analysis of the newly granular CBP dashboard found that $3.26 billion of the $3.94 billion in cumulative UFLPA detentions since June 2022 — 82–83% by value — fell under HTS code 8541, the classification covering solar cells and modules, confirming that solar/polysilicon has been the law's overwhelmingly dominant target despite being formally one of several "high-priority" sectors (Reuters, Feb 24 2026). Detained solar shipments came predominantly from Malaysia, Vietnam, and Thailand — the same third countries that expanded solar manufacturing capacity to serve the U.S. market — rather than from direct China-origin shipments, illustrating how Xinjiang-linked material has moved through intermediate assembly nodes. Roughly 79% of detained solar/electronics value was ultimately released to importers after they demonstrated non-Xinjiang sourcing.

Enforcement intensity has grown even as detained dollar value has fallen: CBP stopped 7,325 shipments in FY2025, a 51% jump over FY2024's 4,850, while the DHS UFLPA Entity List grew from 20 entities at inception to 144 entities by mid-2025 and roughly 116–144 entities through 2026, with dedicated polysilicon/solar sub-categories (LandedFees UFLPA tracker, June 2026; Troutman Pepper, Feb 2026). The shift reflects a pivot from high-value finished solar panels toward lower-value components embedded deeper in supply chains, including microchips and battery materials, as importers successfully diversified panel assembly away from Xinjiang inputs.

US antidumping and countervailing duties on silicon metal and ferrosilicon (2024–2026)

Beyond forced-labor enforcement, the U.S. has layered a separate trade-defence regime directly onto silicon metal and ferrosilicon imports. On ferrosilicon, Commerce finalized combined AD/CVD margins of 1,042% on Russia (finalized Sept. 12, 2024), 14–73% on Brazil, 33–281% on Kazakhstan, and roughly 17–51% on Malaysia (all finalized by Commerce and the ITC in March–April 2025) (Ferroglobe investor presentation, 2026). On silicon metal, Commerce reached final affirmative AD/CVD determinations against Angola (combined 68%), Laos (combined 164%, rising to 204% with reciprocal tariffs), and Thailand (combined 31%, 50% with reciprocal tariffs) on February 18, 2026, followed by Australia (48% combined, 58% with reciprocal tariffs) and Norway (21% combined, 36% with reciprocal tariffs) on June 25, 2026 (Ferroglobe investor presentation, 2026). Ferroglobe, the case petitioner and largest Western merchant silicon-metal producer, states the combined effect of trade cases plus production curtailment should "stabilize supply and demand" and supports a 2026 market upswing (Ferroglobe Q4/FY2025 results release).

EU ferroalloys safeguard: silicon metal excluded, ferroalloys quota-restricted

The European Commission approved EU ferroalloys safeguard measures on November 18, 2025, setting country- and product-specific quotas at 75% of each supplier's average 2022–2024 import volume, with an out-of-quota duty equal to the gap between a reference price threshold and the actual import price (Ferroglobe investor presentation, 2026). Critically, silicon metal itself was excluded from the final safeguard — industry press attributes the exclusion to opposition from EU organic-silicon (silicone) producers, who rely on imported silicon metal as feedstock and lobbied against a measure that would raise their input costs (Qinhuangdao PUDA / Argus commentary, Sept 2025). The exclusion contributed directly to Ferroglobe suspending silicon-metal production in Europe in September 2025, citing market pressures and the absence of the protective measures it had sought (MinExForum, Sept 2025).

Timeline 2020–2026 — from forced-labor law to CHIPS Act to Section 232

A compact chronology of the events that moved silicon — long treated as an abundant, unremarkable commodity — onto the U.S. critical-minerals list and into the center of semiconductor and solar trade policy.

From quartz to ferrosilicon: the feedstock and metallurgical supply chain behind both silicon streams

Silicon metal (~98–99% purity) and ferrosilicon (72–90% Si, balance iron) are both produced by carbothermic reduction of silica — quartz, quartzite, or high-grade silica gravel — with carbon reductants (coal, coke, wood chips, or charcoal) in submerged-arc electric furnaces. Per USGS MCS 2026, U.S. ferrosilicon production is sourced primarily from domestic quartzite, and the domestic depletion allowance schedule explicitly lists quartzite at 14% and gravel at 5% — codifying quartz/quartzite as the tax-recognized ore input for the entire silicon and ferrosilicon industry. USGS also states world and domestic silica resources for silicon-making are "abundant" and "adequate to supply world requirements for many decades," meaning silicon's supply risk sits almost entirely in processing and refining capacity, not raw ore scarcity.

1. Ferrosilicon: the steel industry's deoxidizer and alloying workhorse

Ferrosilicon (typically 50% or 75% silicon grades) is consumed predominantly by the ferrous foundry and steel industries as a deoxidizer and alloying agent, concentrated in the Eastern United States per USGS MCS 2026. Global 2025 ferrosilicon production reached an estimated 75,000 metric tons in the USGS-tabulated "ferrosilicon" line for select producer countries when treated separately from silicon metal (China alone: 3.5 million tonnes in 2025, up from 3.1 million in 2024), with Russia (420,000 t), Norway (150,000 t), and Kazakhstan (120,000 t) rounding out the next tier of producers (USGS MCS 2026, World Production table). USGS's substitutes list for ferrosilicon names aluminum, silicon carbide, and silicomanganese as alternative deoxidizers/alloying agents in some steelmaking applications (USGS MCS 2026). U.S. ferrosilicon import sources for 2021–2024 were dominated by Russia (30%), Brazil (16%), Canada (13%), and Malaysia (11%) — a materially different sourcing mix than silicon metal, and one now heavily reshaped by the Russia AD/CVD order discussed in the trade-flows section above (USGS MCS 2026).

2. Ferroglobe, Elkem, and Wacker: the Western merchant silicon-metal producer base

Ferroglobe PLC describes itself as "the largest merchant producer of silicon metal in the Western World," operating 25 global operations across five continents with 18 electro-metallurgy production centers and more than 50 furnaces, split across Europe (184,000 t silicon-metal capacity), North America (93,200 t), and South Africa (51,000 t) for total silicon-metal capacity of roughly 328,200 tonnes, alongside 302,000 tonnes of silicon-based alloy capacity and 561,500 tonnes of manganese-alloy capacity (Ferroglobe investor presentation, 2026). Ferroglobe reported FY2025 net sales of $1.3 billion (silicon metal $430 million, silicon-based alloys $398 million, manganese alloys $358 million) against adjusted EBITDA of just $28 million, reflecting the severity of the 2025 price trough (Ferroglobe Q4/FY2025 results). Elkem ASA (Norway, historically linked to Wacker and China's Bluestar/ChemChina through joint ventures and past ownership) is a fully integrated global silicones and silicon-materials producer with smelting operations across Norway, Iceland, Canada, and China (Elkem Integrated Annual Report 2024). Wacker Chemie operates its own captive silicon-metal furnaces feeding directly into its polysilicon and silicone chains, most notably in Holla, Norway and at its Charleston, Tennessee complex, giving it upstream security that pure polysilicon refiners lack (Wacker Tennessee overview). Ferroglobe estimates it is only about 40% self-supplied on silicon metal feedstock for its own downstream alloy production, versus 80% self-sufficiency on manganese alloys — illustrating that even the largest Western silicon-metal producer still depends partly on merchant and intercompany purchases to run its full value chain (Ferroglobe investor presentation, 2026).

3. High-purity quartz: Spruce Pine's single point of failure for the entire chip industry

Standard silica feeds ordinary metallurgical silicon and ferrosilicon, but semiconductor-grade polysilicon production depends on an entirely different input: high-purity quartz (HPQ) used to fabricate the crucibles in which polysilicon is melted for wafer growth. The Spruce Pine, North Carolina district hosts pegmatite deposits of exceptionally pure quartz mined by Sibelco (which acquired the legacy Unimin/Iota Quartz operations) and The Quartz Corp, and author Vince Beiser's research is cited as estimating that 70–90% of the crucibles used worldwide to melt polysilicon for computer chips are made from Spruce Pine quartz (U.S. News/AP, Oct 2024). Sibelco markets its Spruce Pine output under the IOTA high-purity quartz brand, explicitly targeting semiconductor, solar, and fiber-optic crucible and reactor-tube applications (Sibelco, High Purity Quartz), having previously announced a major expansion of its Spruce Pine operations to meet growing chip and solar demand (Sibelco, Spruce Pine expansion announcement, Apr 2023). Hurricane Helene forced both Sibelco and The Quartz Corp to shut down Spruce Pine operations on September 26, 2024 as the storm devastated western North Carolina; Sibelco restarted mining and processing just two weeks later, on October 10, 2024, with production and shipments "gradually increasing to full capacity," while The Quartz Corp said existing inventory would prevent any "critical situation" for downstream chip and solar customers (Sibelco, restart announcement, Oct 2024; The Verge, Oct 2024). The episode briefly exposed how concentrated the crucible-quartz supply chain is: unlike bulk silica for metallurgical silicon, which is genuinely globally abundant, semiconductor-grade fused-silica crucible feedstock has effectively two qualified commercial suppliers worldwide, both drawing from the same 50-mile stretch of the Blue Ridge Mountains (Smith Intelligent Distribution, Oct 2024).

4. Aluminum alloying: the largest single demand pool for metallurgical silicon

The single largest end use for metallurgical-grade silicon metal is as an alloying addition for aluminum — primarily aluminum-silicon casting alloys used in automotive engine blocks, wheels, and structural components, where silicon improves castability, reduces shrinkage, and increases wear resistance. USGS confirms aluminum-alloy producers are, alongside the chemical/silicone industry, the main consumers of U.S. silicon metal (USGS MCS 2026), and DOE's Solar Energy Supply Chain Deep Dive similarly frames aluminum alloys as a large, low-growth (roughly 0% CAGR) demand pool alongside faster-growing solar and semiconductor polysilicon and silicones/silanes demand (both modeled around 5% CAGR) (U.S. DOE, Solar Photovoltaics Supply Chain Deep Dive Assessment, 2022). This demand structure means aluminum alloying remains the demand floor beneath silicon-metal pricing even during the depths of the 2025 polysilicon collapse discussed above.

EU strategic-material status, the silicones downstream, and why silicon recycling barely exists

1. EU Critical Raw Materials Act: silicon metal named a Strategic Raw Material

The European Commission's original Critical Raw Materials Act proposal, COM(2023) 160 final of March 16, 2023, lists silicon metal in Annex I among 16 (later 17) Strategic Raw Materials, alongside bismuth, cobalt, gallium, germanium, lithium (battery grade), and rare earths for magnets (European Commission, COM(2023) 160 final, Annex I). Silicon metal also appears in the broader Annex II list of Critical Raw Materials (European Commission, Critical Raw Materials list), and the European Parliament has separately confirmed that under CRMA Article 3, "raw materials, including in unprocessed form, at any stage of processing... shall be considered to be strategic raw materials," with silicon metal listed as a strategic raw material (European Parliament, parliamentary question P-10-2024-002919, Dec 2024). The CRMA's binding 2030 benchmarks — at least 10% of EU strategic-material consumption from EU extraction, 40% from EU processing, 25% from EU recycling, and no more than 65% of EU consumption from any single third country — apply to silicon metal on the same basis as the other 16 Strategic Raw Materials (European Commission, CRMA overview). Notably, when the Commission announced its first tranche of CRMA Strategic Projects in 2025, silicon metal (along with bismuth and titanium metal) was not covered by any Strategic Project in that first list, unlike cobalt, copper, lithium, or rare earth magnet materials (European Commission, Strategic Projects Q&A, 2025). A German government-commissioned analysis further notes that because silicon metal is mined and processed directly (not as a zinc/bauxite byproduct like gallium or germanium), the EU's Industrial Emissions Directive would apply more directly to any EU silicon-metal strategic mining project than it does for byproduct-sourced critical minerals (Umweltbundesamt, CRMA implementation study).

2. EU CBAM: silicon is not yet a covered good, but the exposure is rising

The EU's Carbon Border Adjustment Mechanism, established by Regulation (EU) 2023/956, entered its transitional reporting phase on October 1, 2023 and moves into its definitive phase with certificate purchase obligations from January 1, 2026 (Regulation (EU) 2023/956, consolidated text). CBAM's current scope under Annex I covers six sectors — cement, iron and steel, aluminum, fertilizers, hydrogen, and electricity — identified as the most carbon-intensive and carbon-leakage-exposed EU ETS sectors; silicon metal and ferrosilicon are not among the goods currently listed in CBAM's Annex I (European Commission, CBAM Guidance Document). The free-allowance phase-out that CBAM certificates mirror runs 2.5% in 2026, rising annually to 100% by 2034, per Article 30's phase-in schedule (Climate Leadership Council, Guide to the EU CBAM, 2026). Silicon's CBAM exposure is nonetheless rising indirectly: CBAM's Article 30(3) requires the Commission to assess downstream products at risk of carbon leakage for potential future inclusion, and the Commission's 2025 legislative proposal, COM(2025) 989, explicitly proposes extending CBAM to steel- and aluminum-intensive downstream products (European Parliament, COM(2025) 989) — a category that could eventually sweep in aluminum-silicon casting alloys, since silicon is a structural additive to the aluminum products CBAM already covers, even though silicon metal itself remains outside CBAM's current Annex I in July 2026.

3. Silicones: the largest non-metal, non-solar demand pool for silicon metal

USGS confirms the chemical industry — "in particular for the manufacture of silicones" — is one of the two main consumers of silicon metal alongside aluminum alloys (USGS MCS 2026). Silicones (organosilicon polymers) are manufactured by reacting silicon metal with methyl chloride in the "Direct Process" to make chlorosilanes, then hydrolyzing and polymerizing them into silicone fluids, elastomers, and resins used across construction sealants, personal care, electronics encapsulation, and medical devices. The global silicone market was valued at roughly $18.2–24.5 billion in 2024–2025, with a five-player group — Wacker Chemie, Shin-Etsu Chemical, Elkem, Dow (successor to the Dow Corning joint venture), and Momentive Performance Materials — together holding roughly 70–75% of global capacity (MarketsandMarkets, Silicone Market analysis; DataHorizzon Research, Silicones Market, 2025). Dow and Wacker compete substantially on backward integration, owning their own silicon-metal furnaces and chlorosilane plants, while Momentive (majority owned since 2024 by South Korea's KCC Corporation, which completed its acquisition of Momentive in May 2024) and Shin-Etsu differentiate more through specialty formulation chemistry and applications engineering (Research and Markets, Silicone Market Report 2026). The historical Dow Corning joint venture (Dow Chemical/Corning Inc.) was restructured into Dow's wholly owned Dow Silicones business after Dow acquired full ownership in 2016, and Dow continues to expand downstream silicones capacity in China, including a March 2025 completion of a Zhangjiagang Bonded Zone expansion project (MarketsandMarkets, Silicone Market Report). Elkem, historically linked to China's Bluestar (ChemChina) group through joint-venture and ownership history in the silicones space, in 2026 restructured its own silicones portfolio via a divestment to Bluestar, underscoring the continuing consolidation of Western/Chinese silicone capacity even as the upstream silicon-metal market remains oversupplied (USD Analytics, Silicone Resins Market press release, 2026).

4. Recycling and substitution: both structurally weak for silicon

USGS states plainly that silicon recycling is "insignificant" (USGS MCS 2026) — there is no dedicated silicon-metal scrap-collection industry comparable to aluminum or steel scrap. The one meaningful recycling pathway runs indirectly through aluminum scrap: because silicon is alloyed directly into cast aluminum, aluminum recycling (which the International Aluminium Institute puts at a 76% global recycling efficiency rate) inherently recovers and recirculates the silicon content of aluminum-silicon alloys along with the aluminum itself, without any separate silicon-specific recovery step (International Aluminium Institute data, via Wikipedia summary). Academic research has explored dedicated silicon recovery from aluminum die-casting scrap for solar feedstock, achieving purities up to 99.96% in laboratory trials, and separately from silicon wafer-cutting waste, achieving silicon recovery yields near 96% using cryolite-assisted alloying — but neither pathway has scaled into a commercial industrial-recycling stream (Journal of Crystal Growth, Al–Si die-casting scrap recycling study; Science of the Total Environment, silicon cutting-waste recycling study). End-of-life solar panels — the fastest-growing potential silicon waste stream given the multi-decade global PV build-out — have essentially no established silicon-recovery infrastructure at scale as of 2026. On substitution, USGS confirms silicon has almost no viable replacement in its highest-value uses: only gallium arsenide and germanium serve as substitutes in semiconductor and infrared applications, and only in narrow, high-cost niches, while aluminum, silicon carbide, and silicomanganese can substitute for ferrosilicon in some steelmaking applications (USGS MCS 2026). For solar photovoltaics specifically, crystalline silicon remains the dominant commercial technology; DOE's Solar Futures Study (2021) envisions cumulative U.S. solar deployment rising from about 80 GW in 2020 to 760–1,000 GW by 2035 and 1,050–1,570 GW by 2050 under grid-decarbonization scenarios, with R&D on thin-film and perovskite-silicon tandem cells flagged as the main avenue to reduce (not eliminate) silicon intensity per watt, rather than to substitute away from silicon altogether (OSTI, Policy-Driven Solar Innovation commentary on the DOE Solar Futures Study).