Primary source → BIS Applied Materials settlement documents, 11 February 2026
If you run a Korean company that buys U.S.-origin parts, finishes a product in Korea, and ships it to a Chinese customer, there is one sentence from the February 2026 Applied Materials settlement you need to read before anything else:
“‘substantial transformation’ does not appear anywhere in the EAR and is not the correct test for determining whether an item is subject to the EAR because it is an item of U.S. origin.” (Order, p. 11)
That sentence, written by the U.S. Bureau of Industry and Security, dismantles the most common assumption in the Korean semiconductor supply chain — the belief that doing meaningful work on a product in Korea makes it “Korean” for U.S. export purposes. It does not. And the price of learning that the hard way, in this case, was $252,500,300.
What actually happened
Applied Materials partially built ion-implant tools in Gloucester, Massachusetts. When SMIC — the Chinese chipmaker — placed an order, AMAT shipped the partially assembled tools, plus all the U.S.- and foreign-origin parts needed to finish them, to its Korean subsidiary in Pyeongtaek. AMK completed assembly and testing, then shipped the finished tools to SMIC in China (Order, p. 8).
The problem: SMIC was subject to a U.S. license requirement for these tools from 25 September 2020, and was placed on the Entity List — requiring a license for all U.S.-regulated items — on 18 December 2020 (Order, p. 4). The ion-implant equipment reexported to SMIC was valued at about $126,250,150 and classified under ECCN 3B991 (Order, p. 2). On 11 February 2026, AMAT and AMK settled 54 charges for the statutory-maximum penalty, plus a three-year suspended denial order and two years of mandatory compliance audits.
Note the classification: 3B991 is a relatively low-tier commercial item, not a weapon and not high-end controlled technology. Many Korean compliance teams assume EAR extraterritorial reach only bites for top-tier dual-use goods. A $252.5M case built on 3B991 equipment finished in Korea says otherwise.
Why “Made in Korea” did not help
The instinct — borrowed from customs and country-of-origin rules — is that enough work in Korea changes the product’s origin. That logic governs tariffs. It does not govern U.S. export jurisdiction, which turns on a different provision (15 C.F.R. § 734.3): U.S.-origin items remain subject to the EAR wherever they go. Finishing them abroad does not reset that status.
BIS was blunt about the mechanics. In a footnote absent from every news account, it pointed to exactly where AMAT’s analysis failed: a U.S.-origin item should skip the “de minimis” calculation entirely and go straight to the license question; AMAT ran the de minimis step it was never entitled to run (Order, pp. 11–12, fn. 15). And it set out a standard worth memorizing — U.S.-origin items are not made “foreign” by overseas assembly when those activities involve “little or no foreign-origin parts… shipped to the foreign location from a non-U.S. location” (Order, p. 12).
There is a flip side that is genuinely useful. BIS excluded the Singapore-origin enclosure and factory interface from liability because they were truly foreign-made and cleared the de minimis and foreign-direct-product tests (Order, p. 7). That gives you a boundary you can act on: genuinely foreign-made sub-assemblies that clear those tests are not your exposure. The U.S.-origin content you import and integrate is.
The part that should worry a Korean SME
Read the production model again as a structural diagram, not a story about one American company: U.S. partial build → Korean completion → China delivery. That is how a large share of Korea’s equipment supply chain operates with U.S.-origin subsystems — vacuum chambers, RF generators, controllers, gas-delivery modules — sourced from U.S. OEMs, finished in Korea, and shipped to Chinese fabs. If you can map your own bill of materials onto AMAT’s sentence, you share AMAT’s exposure.
Two further details make the analogy uncomfortably close. AMK was a refurbishment facility before 2021, repurposed into an assembly hub to serve the China workaround (Order, p. 3). If your U.S. parent or customer has recently asked you to expand assembly scope for China-bound orders, that is the same move. And BIS was explicit that the offshoring structure itself is the violation — the analysis “did not change when, in order to continue selling to a single customer on the Entity List, AMAT… partially moved assembly and testing activities outside the United States for specific items… ordered by that single customer” (Order, p. 11).
Even the motive is on the record. Losing SMIC meant losing “$112–150 million in annual revenue,” and losing all of SMIC’s business meant “more than $1 billion per year” (Order, p. 10). BIS knew about the commercial pressure and enforced anyway. If your China revenue is 30–50% of the total, that dependency is not your defense — it is precisely what BIS documented as the reason the violation happened.
What a mature compliance program did not buy
Here is the detail that should end any “we have a program, we’re fine” complacency: AMAT had “designed, implemented, and maintained an export compliance program” and had “applied for over 1,100 licenses from BIS,” including over 100 for SMIC-related shipments (Order, p. 2). It still paid the statutory maximum. Good-faith licensing on parallel product lines did not offset a deliberate structural workaround on the dual-build stream.
What to do this quarter
You do not need to wait for your own order to act. Three moves cost almost nothing:
- Pull every EAR-scope opinion that relies on Korean assembly to “reset” U.S.-origin status, and re-open it. The substantial-transformation logic is dead for this purpose.
- Segment your BOM against the AMAT boundary. Foreign-made sub-assemblies that clear de minimis and FDPR are outside the EAR; U.S.-origin content you import and integrate is inside it. That single line gives you a fast exposure map.
- Treat any “is-informed” letter from BIS as a step-change. It moved AMAT’s license universe from one ECCN to everything in under three months (Order, p. 4).
The AMAT case is not really about Applied Materials. It is the most precise public statement BIS has made about how its jurisdiction follows U.S.-origin content into a Korean factory and back out again to China. The companies that read it as a case study about someone else are the ones most likely to be its next chapter.
This is the lead piece in a series on the AMAT settlement and Korea–China export-control exposure. The full long-form analysis covers the jurisdictional theory and audit timeline in depth; companion pieces cover the BIS warning-letter mechanics, the Cadence EDA precedent, and what Chinese buyers should ask their Korean suppliers. A Korean-language version follows on Brunch, adapted for Korean SME readers.