Taiwan vs China Sourcing: Tariffs and Origin

Why there are no percentages in this article

If you buy hardware for the US market, you have watched tariff rates on both Taiwan and China move repeatedly since 2025: emergency economic authorities invoked and litigated, a Supreme Court decision, statutory fallbacks with sunset dates, and bilateral agreements layered on top. Individual rates have been announced, superseded, struck down and renegotiated inside the same quarter.

So this article quotes none of them. A number printed here would be wrong before most readers found the page, and a sourcing decision built on a stale rate is worse than one built on no rate at all.

What has been stable through all of it is the structure: three things decide what you pay at the border, and only one of them is the headline rate.

What decides your duty	Who controls it
The HTS classification of the finished product	Product design and how it is described
The country of origin	Where meaningful manufacturing happens, not where it ships from
Which tariff layers attach to that origin	Trade policy, which is the part that keeps moving

Two of those three are decided in engineering and sourcing, long before anything reaches a port. That is the part worth your attention, because it is the part you can act on.

Country of origin is not where you assemble

This is the most expensive misunderstanding in the category.

US Customs and Border Protection determines origin using a substantial transformation test: an article acquires a new origin when processing gives it a new name, character, or use. The corollary matters more than the rule. Assembly that is minimal or simple does not confer origin. A product built from finished modules with a screwdriver, in a country those modules were shipped to, generally keeps the origin of the modules.

So a device assembled in Taiwan can be Taiwanese origin, or it can be Chinese origin that passed through Taiwan. The difference is not the address on the packing list. It is how much of the product’s identity was created there.

CBP decides this case by case, on the totality of the circumstances. That sounds unhelpfully vague until you realize what it implies: the answer depends on your bill of materials and your process, which means it is a design question, and design questions can be answered before you commit.

Origin follows the bill of materials

Once you accept that origin tracks meaningful manufacturing, the sourcing decision changes shape. The question stops being “where do we do final assembly” and becomes:

• What arrives as a finished functional module, and what arrives as raw material or components? A populated board that only needs to be screwed into a housing is a very different origin fact than bare boards, discrete components and a local SMT line.
• Where does the electronic transformation happen? For most connected products, the moment a bare PCB becomes a working assembly, through solder paste, placement and reflow, is the strongest candidate for where the product acquired its character.
• What proportion of value and processing is local? On its own it does not settle the question, though CBP weighs it as part of the totality.

None of these are logistics questions. They are decisions about where the supply chain sits, made when the BOM is set. A sourcing strategy that treats final assembly as the lever is optimizing the one variable that carries the least weight.

→ How a bill of materials gets graded for risk before tooling: Component Supply Risk Matrix for OEM Projects

You can find out before you commit

Because substantial transformation is decided case by case, brands often treat it as unknowable until the goods arrive. It is not.

CBP publishes its rulings in a searchable database, and prior rulings on comparable products show how the agency has reasoned about similar facts. For a program where the stakes justify it, an importer can also request a binding ruling on country of origin before importing, describing the actual manufacturing process and receiving a determination that customs will honor.

That is a very different posture from finding out at the port. It also changes what you should be asking a prospective manufacturer: not just “can you build this,” but “can you describe your process in enough detail to support an origin determination, and will you put that in writing?”

A factory that cannot describe which operations happen under its own roof, and which arrive as finished assemblies from elsewhere, cannot help you answer the question that decides your duty rate.

The second exposure: forced-labor enforcement

Tariffs are not the only border risk attached to a supply chain, and the second one works differently enough that treating it as a tariff problem gets it wrong.

Under US forced-labor law, goods connected to listed entities or regions face a rebuttable presumption of inadmissibility. Three features make this a different kind of problem:

1. The burden sits entirely with the importer of record, which is usually the brand, not the factory.
2. The standard is clear and convincing evidence, higher than the ordinary civil standard.
3. Rebutting it requires tracing the supply chain back through every tier to raw inputs, with sourcing records, transport documentation, employment records and independent audits.

Enforcement has expanded steadily, and the entity list has grown through 2026 to cover electronics, metals and battery materials among other sectors. Lithium cells and components containing certain minerals draw particular attention, which puts most rechargeable consumer hardware inside the scope of the question.

The number that matters here is not the duty rate. It is that only a small minority of detained shipments are successfully released, because assembling clear and convincing evidence after the fact is extremely difficult.

That makes this a documentation problem, and the documentation lives with your manufacturer. Whether the traceability records exist at all is determined by how the factory buys and records materials, years before anyone needs them.

So what actually differs

The honest framing is not that one origin is good and another is bad. It is that the two carry different structures of risk, and the structures do not trade off against each other in a single number:

• Duty exposure depends on origin, and origin depends on how much of the product is really made where you think it is made.
• Forced-labor exposure depends on documentation depth through every tier, and it lands on you rather than on your supplier.
• Both are decided by decisions taken during development, and both are extremely expensive to fix after the product ships.

A unit price comparison captures none of this. It is the number that is easiest to obtain and the least predictive over a product’s life, which is why it tends to dominate early conversations and disappoint later ones.

→ The wider evaluation framework, beyond duty and unit cost: Taiwan as a China+1 for Health Devices

What to ask, before rates change again

Three questions that stay useful regardless of what the rate is this quarter:

1. Which operations happen under your roof, and which arrive finished? Ask for it operation by operation, not as a summary.
2. Can you support an origin determination with process documentation? And if the answer is yes, will it survive being shown to customs?
3. How far back can you trace materials, and what records exist? Not “do you comply,” but “what documents would we actually hand over.”

A note on scope

This is general information about how origin and border enforcement work, not legal, customs or trade advice, and the rates and entity lists referenced move frequently. Verify current rates for your specific HTS classification and confirm origin questions with a licensed customs broker or trade counsel before making a sourcing commitment. What is durable here is the structure, not any number.

→ The manufacturing side of these questions, from schematic to SMT line: Consumer Electronics Contract Manufacturing