Does Moving Your Sourcing Actually Cut Your Tariff Bill? Which 2026 Duties Follow Origin — and Which Follow the Metal

Does moving your sourcing out of China actually cut your tariff bill in 2026?

Partly — it cuts the tariff layers that follow country of origin, but not the ones that follow the material. This is the distinction that decides whether a multimillion-dollar supply-chain move pays for itself. Section 301 (25% on Chinese-origin goods) and the base MFN rate are origin-based, so a genuine substantial transformation in Vietnam, Mexico, or India removes them. Section 232 on steel, aluminum, and copper is not: its exemptions and penalty rates track where the metal was melted-and-poured or smelted-and-cast, so moving where the finished article is assembled does not shed it. For metal-intensive and capital equipment now dutied on full customs value, that locked layer can be the majority of the stack — which is why the honest answer to "should we relocate to escape the tariff" starts with "which tariff?"

The frame that matters: Your duty stack is two ledgers, not one. Origin-based duty (301, base MFN) is escapable — it moves when your country of origin genuinely moves. Material-based duty (232) is locked — it follows the furnace, not the factory. A sourcing shift can only ever recover the escapable ledger, and on heavy equipment that is the smaller one.

This post is the sourcing-decision companion to the rest of the 2026 duty cluster: the capital-goods import surge, why your duty is outrunning your volume, and the duty leakage recovery guide. It is a reality check, not a how-to — for the mechanics of executing an origin change, this links you to the right playbooks rather than repeating them. Last updated: July 2026.

Which duties follow origin, and which follow the metal?

Split your stack into two ledgers before you cost a single relocation. Every layer of a 2026 duty bill is either origin-based (it moves when your country of origin genuinely moves) or material-based (it doesn't). Here is where the common layers fall:

Duty layer	What it follows	Escapable by a sourcing shift?
Base MFN / Column 1 rate	Country of origin (substantial transformation), reducible via FTA	Yes — genuine origin change or FTA qualification
Section 301 (China, 25% on Ch 84/85)	Country of origin (substantial transformation)	Yes — genuine substantial transformation in a third country
Section 232 steel	Country of melt-and-pour	No — follows where the steel was made
Section 232 aluminum	Country of smelt-and-cast	No — follows where the aluminum was made
Section 232 copper (from July 30, 2026)	Country of smelt-and-cast	No — follows where the copper was made
AD/CVD (if an order covers your goods)	Scope + origin, heavily enforced	Rarely, and risky — see the enforcement section

The escapable ledger is real money — for a Chinese-origin line, the 25% Section 301 layer plus the base rate is often the largest single thing a sourcing move can remove. But the moment steel, aluminum, or copper content is involved, a second ledger opens that a change of assembly country does not touch. For how origin is actually determined, see country of origin determination rules; for the 301 playbook, see Section 301 strategy for Chinese-origin imports.

Why moving the factory doesn't move the furnace (Section 232 melt-and-pour)

Section 232 has its own origin test, and it is the reason a sourcing shift hits a wall on metal goods. Country of origin under the traditional substantial transformation test — the "name, character, and use" standard — sets origin for Section 301, Section 201, and marking. But Section 232 layers a second rule on top: for steel, the duty exemptions and penalties track the country of melt and pour — "the original location where the raw steel is first produced in a steel-making furnace in a liquid state and then poured into its first solid shape" (CBP CSMS #62582900). For aluminum, it is the country of smelt and cast (CBP CSMS #55424218); for copper, the same smelt-and-cast basis takes effect July 30, 2026 (CBP CSMS #69252300).

The government closed the obvious workaround deliberately. As trade counsel summarized CBP's guidance on Proclamation 10783:

"Previously, steel that was melted and poured in a non-exempted country could have been imported into Mexico, undergone further processing in Mexico to effectuate a substantial transformation, and had Mexico as the country of origin to avoid Section 232 duties. Proclamation 10783 thwarts this course of action by requiring proof that the steel article was melted and poured in a North American territory in order to obtain the benefit of not paying Section 232 duties."

So you can substantially transform a metal article abroad and legitimately change its origin for 301 — and still owe the 232 metal duty, because the furnace didn't move. Where the melt-and-pour or smelt-and-cast country can't be established, the consequences get worse, not better: unknown-origin metal is reported "OTH" or "UN" and can face duties up to 200% (CBP CSMS #65340246). Moving the factory doesn't move the furnace — it just adds a documentation burden.

The full-value amplifier: why the locked layer got bigger in 2026

The locked ledger isn't just unavoidable in 2026 — it's larger than it has ever been, because Section 232 now applies to the full customs value of the finished article. Under Proclamation 11021 (issued April 2, 2026; effective April 6), Section 232 duties on steel, aluminum, copper, and their derivatives apply to the entire customs value of the imported product, regardless of the actual value of the embedded metal — a change from the prior practice of taxing metal content only. Rates run to 50% on most covered articles, 25% on certain derivatives, and 10% for derivatives made with US-melted metal, with a temporary 15% track for certain Chapter 84/85/87 equipment through December 31, 2027.

Two consequences for a sourcing decision:

• The ceiling on savings is lower than it looks. When 232 was a duty on metal content, a metal-light finished good carried little 232. Now a $2 million machine with modest steel content can owe 232 on the whole $2 million. The locked ledger grew; the escapable share shrank.
• One metal, one duty — but the biggest one. Where a product is a derivative of more than one covered metal, only one 232 duty applies (no metals-on-metals stacking) — but it is the highest applicable rate, on the full value. There is no averaging your way out of it.

This is the same force described in how tariff stacking works and the reason your duty can outrun your volume: the layers that grew in 2026 are largely the ones a sourcing shift can't reach.

Where did importers actually move their sourcing in 2026?

The shift out of China is real, large, and already in the trade data — the question is whether it moved the duty. US goods imports from China fell to $308.4 billion in 2025, down $130.4 billion (about 29.7%) from 2024 (US BEA and Census Bureau, Annual 2025 release, February 19, 2026). The volume didn't vanish; it relocated. Mexico held the number-one spot at $534.9 billion, and Vietnam set a record at $193.8 billion, up $57.3 billion year over year.

The cleanest illustration of escapable-versus-locked is smartphones. India's share of US smartphone imports jumped from roughly 13% to 44% in the year to mid-2025, while China's fell from about 61% to 25% (Canalys). That shift worked cleanly on the tariff because a smartphone is low-metal and genuinely transformed in India — the origin moved, Section 301 fell away, and there is no meaningful 232 metal ceiling on a phone. Heavy electrical and mechanical equipment is the mirror image: the origin can move, but the copper in the windings and the steel in the frame keep their 232 duty. The lesson isn't "don't move" — it's that the payoff depends on your metal intensity, which is exactly what a phone maker and a transformer importer get opposite answers to.

How to calculate what a sourcing shift will actually save

Don't estimate the saving as "the tariff" — estimate it as the escapable share, net of switching cost. The arithmetic is simple once the stack is decomposed:

1. Split each product's landed duty into origin-based (301 + base MFN) and material-based (232) dollars.
2. The escapable ceiling is the origin-based dollars — and only if a genuine substantial transformation or FTA qualification is achievable at the new origin.
3. Subtract the switching cost: origin-qualification work, retooling, typically higher unit cost at the new source, and the ongoing EAPA/documentation burden of defending the origin claim.
4. What's left is the real, defensible saving. For a metal-light finished good it can be most of the tariff; for capital equipment it is often a minority of it.

You can size a single lane with the tariff calculator or a landed-cost comparison across sourcing countries, and you should run it before signing a relocation, not after. A move that recovers a 25% Section 301 layer on a metal-light product is often a clear win; the same move on a full-value-232 machine can cost more to execute than it saves.

The enforcement trap: origin follows the input, not the label

The line between legal diversification and illegal evasion is whether the manufacturing actually moved — and CBP is enforcing it hard. Under the Enforce and Protect Act (EAPA), CBP's public statistics show well over 200 investigations since 2016, a substantial share involving transshipment. The determinations are pointed. In the Waaree solar case (EAPA determination, June 23, 2026), CBP found evasion where photovoltaic modules were assembled in India but the dutiable input — the solar cell — was subject to the AD/CVD orders, holding that for AD/CVD purposes origin follows the cell, not the module — and it found evasion despite the importer's full cooperation and on-site verification access. (CBP separately confirmed on verification that no Chinese-origin cells were used; the evasion was declaring India, the country of assembly, as the origin.)

The principle rhymes with melt-and-pour: origin can follow the dutiable input, not the last assembly step — though AD/CVD scope and common-law substantial transformation are distinct tests. The related substantial-transformation rule the courts apply is that mere assembly or minor processing is not enough to change origin (Energizer Battery, Inc. v. United States). A genuine shift of real manufacturing is lawful tariff planning; a sham shift — relabeling or trivial processing to disguise the true origin — is evasion under 19 U.S.C. 1592, with civil penalties stacked on top of the back duty. If you are moving origin, the documentation that proves the transformation was real is not paperwork overhead; it is the entire defense. For the AD/CVD dimension, see the antidumping and countervailing duties guide and the AD/CVD scope and EAPA evasion defense program.

What a sourcing decision should model this quarter

1. Decompose the stack on your top-value lines into escapable (301 + base) and locked (232) dollars. The ratio is your sourcing-shift ceiling.
2. Test whether a genuine origin change is achievable at the candidate country — real substantial transformation or FTA qualification, not assembly relabeling. If it isn't genuine, it isn't a plan.
3. Price the switching cost — qualification, retooling, unit-cost delta, and EAPA documentation — and net it against the escapable ceiling.
4. Separate the metal-light wins from the metal-heavy traps. Move the products where origin does the work; for capital equipment, pair the move with classification and recovery levers that reach the locked layer.
5. Document the transformation as you go, because the origin claim has to survive a CF-28 and, potentially, an EAPA investigation.

How GingerControl fits — honestly

GingerControl is AI trade-compliance and duty-recovery infrastructure for enterprise and mid-market importers. On a sourcing decision, we do four things: model the full 2026 duty stack per HTS code and origin, decompose it into the escapable layers (301, base MFN) and the locked layer (232 melt-and-pour / smelt-and-cast) so you can see the real ceiling on a move, quantify what a shift could realistically recover net of switching cost, and produce the substantial-transformation and classification documentation the origin position must survive on.

What we do not do: we are not a licensed customs broker, we do not file entries, and we are not your importer of record — your broker stays your broker. We never guarantee a saving or CBP acceptance of an origin position, and we will tell you when the locked 232 layer means a relocation won't pay for itself. The one filing we handle directly is the IEEPA refund package for 2025 duties, which is distinct from filing customs entries. Two-sided honesty is the point: an origin position that can't survive an EAPA inquiry is worse than no move at all.

For the enterprise-scale version, see the enterprise importers solution, or start with a fixed-scope Two-Week Assessment. If your move is specifically China-to-Mexico, the nearshoring and USMCA qualification guide is the how-to companion to this decision frame.

Frequently Asked Questions

Does moving our sourcing out of China actually cut our tariff bill in 2026?

Partly — it cuts the layers that follow country of origin, not the ones that follow the material. Section 301 (25% on Chinese-origin goods) and the base MFN rate are origin-based, so a genuine substantial transformation in Vietnam, Mexico, or India removes them. Section 232 on steel, aluminum, and copper is not origin-based in the same way: it follows where the metal was melted-and-poured or smelted-and-cast, so assembling the finished article somewhere else does not shed it. For a China-heavy line running 301 plus full-value 232, moving origin might recover the 25% and the base rate while leaving the 232 layer intact. The honest number is the escapable share of your stack, net of the cost to switch — not the whole tariff.

Can we escape Section 232 steel, aluminum, or copper tariffs by changing country of origin?

Generally no. Section 232's exemptions and penalty rates are keyed to the country where the metal was melted-and-poured (steel) or smelted-and-cast (aluminum and, from July 30, 2026, copper) — not where the finished article was assembled. Proclamation 10783 in 2024 specifically closed the loophole where steel melted in a non-exempt country could be finished in Mexico to claim Mexican origin and dodge 232. So you can substantially transform a metal article in a third country and change its origin for 301 purposes, yet still owe the 232 metal duty because the furnace didn't move. On a $2 million machine now dutied on full customs value, that locked layer can be the majority of the tariff.

What's the difference between substantial-transformation origin and Section 232 melt-and-pour origin?

They are two different tests, and 2026 tariff planning lives in the gap between them. Substantial transformation — the traditional "name, character, and use" test — sets a good's country of origin for Section 301, Section 201, and marking; it changes when real manufacturing (not mere assembly) occurs in a new country. Section 232 adds an overlay on top: for steel, aluminum, and copper, the exemption and penalty structure tracks the country of melt-and-pour or smelt-and-cast, a separate fact about where the metal was born. A product can be "made in Vietnam" for origin and still carry a "melted in China" steel duty. Escapable duty rides on the first test; locked duty rides on the second.

Does nearshoring to Mexico eliminate Section 301?

It can, if the product genuinely qualifies — but qualifying is the hard part, and it does not touch Section 232. A genuine substantial transformation in Mexico (or USMCA qualification under the product-specific tariff-shift rules) can change origin and remove the 25% Section 301 layer. What it does not do: it does not remove Section 232 on the steel, aluminum, or copper content, because the 2024 melt-and-pour rule requires North American melt-and-pour to earn the 232 exemption. So a Mexican-origin machine built from Chinese-melted steel escapes 301 and still owes 232. The mechanics of qualifying are in our China-to-Mexico USMCA guide; the point here is to model both layers before assuming relocation zeroes the tariff.

Where did US importers actually move their sourcing in 2026?

The shift is real and large. US goods imports from China fell to $308.4 billion in 2025, down $130.4 billion — about 29.7% — from 2024 (BEA and Census, Annual 2025 release, February 19, 2026). Mexico stayed the number-one source at $534.9 billion, and Vietnam set a record at $193.8 billion, up $57.3 billion. The clearest single case is smartphones: India's share of US smartphone imports jumped from roughly 13% to 44% in the year to mid-2025 while China's fell from about 61% to 25% (Canalys). Phones are low-metal and the transformation is real, so origin shifted cleanly — 301 escaped, and there is no 232 metal ceiling on a phone. Heavy equipment is the opposite case.

Isn't a fast origin switch a transshipment or EAPA risk?

Yes, if the switch is a label rather than real manufacturing. CBP enforces country-of-origin claims aggressively under the Enforce and Protect Act: its public statistics show well over 200 investigations since 2016, a substantial share involving transshipment. In a June 23, 2026 determination (the Waaree solar case), CBP found evasion where modules were assembled in India but the dutiable input — the solar cell — was subject to the AD/CVD orders, holding that for AD/CVD purposes "origin follows the cell, not the module," and did so despite the importer's full cooperation. Simple assembly or minor processing is not substantial transformation (Energizer Battery v. United States); a sham shift is evasion under 19 U.S.C. 1592, with penalties on top of the back duty. Legal diversification moves the real manufacturing; illegal diversification moves a label.

How do we figure out which of our duty is escapable before relocating?

Decompose the stack line by line. For each product, split the landed duty into the origin-based layers (Section 301 and the base MFN rate, both escapable by a genuine origin change) and the material-based layer (Section 232, locked to melt-and-pour or smelt-and-cast). The escapable share tells you the ceiling on what a sourcing move can save; subtract the switching cost — qualification, retooling, higher unit cost, EAPA documentation — to get the net. GingerControl models the full 2026 stack per HTS code and origin, separates the escapable layers from the locked ones, and produces the origin and classification documentation the position rests on. We do not guarantee a saving, and we do not file your entries — your broker does.

Does GingerControl file our entries or guarantee the savings?

No. GingerControl is not a licensed customs broker, does not file entries, and is not your importer of record — your broker stays your broker. We model the full 2026 duty stack per HTS code and origin, decompose it into the escapable (301, base) and locked (232) layers, quantify what a sourcing shift could realistically recover, and produce the substantial-transformation and classification documentation the position must survive on. We never guarantee a saving or CBP acceptance of any origin position. The one filing we handle directly is the IEEPA refund package for 2025 duties, which is distinct from filing customs entries.

---

Model What's Escapable Before You Relocate

A sourcing shift only ever recovers the origin-based half of your stack, and on metal-intensive goods that half is smaller than it looks. Before you move a supply chain, GingerControl decomposes your 2026 duty into the escapable layers and the locked 232 layer, quantifies the realistic net saving, and produces the origin documentation the position has to survive. Book a no-cost consultation with Chen to size what's actually escapable.

---

Related Articles

• Capital-Goods Imports Hit $128B in May 2026: What the Surge Means for Your Duty Bill
• Why Your Duty Bill Is Outpacing Your Import Volume in 2026
• Country of Origin Determination Rules: Substantial Transformation
• China-to-Mexico Nearshoring & USMCA Qualification
• Section 301 Strategy for Chinese-Origin Imports
• Tariff Engineering: Legal Strategies to Reduce Import Duty
• Section 232 Steel & Aluminum Tariffs
• The Duty Leakage Recovery Program: From First Finding to Filed Claim

---

• Material Composition Tariff Impact: Designer''s HTS Duty Guide

References

[REF 1] CBP, Section 232 steel country of melt and pour — CSMS #62582900 (reporting effective November 21, 2024). Aluminum smelt-and-cast — CSMS #55424218 (effective April 10, 2023); "UN" unknown origin 200% duty — CSMS #65340246 (effective June 28, 2025). Copper smelt-and-cast — CSMS #69252300 (effective July 30, 2026). Source: CBP CSMS bulletins

[REF 2] Section 232 full customs value on steel/aluminum/copper derivatives — Proclamation 11021 (issued April 2, 2026; effective April 6, 2026); CBP CSMS #68253075. Rates 50% / 25% / 10% (derivatives of US-melted metal) / temporary 15% for certain Ch 84/85/87 equipment through Dec 31, 2027; one-metal-one-duty (no metals-on-metals stacking). Source: Foley & Lardner, Section 232 derivatives | BDO

[REF 3] Proclamation 10783 (2024) — melt-and-pour requirement closing the substantial-transformation-in-Mexico workaround for Section 232 steel. Source: Barnes, Richardson & Colburn, CBP guidance on steel derivative duties

[REF 4] Substantial transformation standard ("name, character, and use"); assembly of foreign components is generally not a substantial transformation — Energizer Battery, Inc. v. United States, 190 F. Supp. 3d 1308 (Ct. Int'l Trade 2016). Source: Steptoe analysis of the substantial-transformation standard

[REF 5] 2026 sourcing-shift data — US goods imports from China $308.4B (−$130.4B, −29.7%) in 2025; Mexico $534.9B (#1); Vietnam $193.8B record. US BEA & Census Bureau, U.S. International Trade in Goods and Services, December and Annual 2025 (released February 19, 2026). India/China smartphone import-share swing per Canalys (mid-2025). Source: BEA Annual 2025 release

[REF 6] EAPA transshipment enforcement (200+ investigations since 2016, a substantial share transshipment); Waaree solar determination (June 23, 2026) — for AD/CVD purposes, origin follows the solar cell, not the module. Sham origin shift penalized under 19 U.S.C. 1592. Source: CBP EAPA statistics