President Trump signed a Section 232 proclamation on 6 August imposing a 15% ad valorem duty on imports of polysilicon, polysilicon ingots and polysilicon derivatives, regardless of country of origin. It takes effect at 12:01 a.m. ET on 4 December 2026 — 120 days out.
Floors, not just tariffs
The unusual instrument here is the minimum import price. The proclamation sets $21/kg for polysilicon, $100/kg for ingots and wafers, $0.22/W for solar cells and $0.38/W for modules. A tariff scales with the invoice; a price floor removes the option of undercutting it.
Who pays what
EU member states, Japan, Korea, Taiwan, Switzerland and Liechtenstein face a combined tariff cap of 15%; the UK rate is 10%. Several major suppliers therefore do not face a straight 15% stack. The order also carries manufacturing drawback, foreign-trade-zone restrictions and an onshoring relief programme for firms building US production, with a construction deadline of 20 January 2029.
The AI framing is the government's own
The proclamation's justification cites polysilicon as essential to national security and to supporting artificial-intelligence innovation. That is the administration's language, and it is the reason the measure is being read as chip policy at all.
What it is not
This is an import tariff, not an export control, and the binding measures cover polysilicon and the solar supply chain. Semiconductor wafers begin as polysilicon, which is the real link to data-centre buildout — but headlines calling it a "chip tariff" describe a scope the document does not have. Nothing in it touches finished accelerators, packaging or memory, and the 120-day runway gives importers a quarter to pull orders forward before any of it binds.