The tariff arithmetic says Vietnam. The shipping manifests increasingly say China.
China's effective US tariff runs about 20%, against 6.1% for Vietnam, 13.4% for Indonesia and 4.5% for Thailand, per Economist Intelligence Unit estimates from July. On paper, that is a 14-point subsidy for moving the factory.
On the ground, it is not enough. Last week the tariff-law collapse stripped the legal foundation out from under a chunk of the administration's schedule, and the Sept 9 policy whiplash told everyone what comes next. This week the operative questions stopped being "what rate applies" and became "who is enforcing it, under what authority, and for how long."
This week the operative questions stopped being 'what rate applies' and became 'who is enforcing it, under what authority, and for how long.'
Start with one reversed shipment. Jin Chaofeng, an outdoor-furniture exporter in Hangzhou, opened a workshop in Ho Chi Minh City in 2024 and shut it this year, moving production home.
His reason wasn't tariffs. He couldn't find the equipment he needed in Vietnam. He was importing basic items from China anyway — screws, and the moulds for cup holders.
"Once I factored everything in, the overall cost was not much different, so there was no point," he said.
That is the whole story in one sentence. The tariff gap was real; the ecosystem gap ate it.
Dawang Metals, a family-owned metal casting company in Dandong, watched a major US agricultural machinery customer shift orders to India. The customer came back. It had run into problems there.
Dawang had also explored moving offshore before abandoning the plan. "China's supply-chain advantage is still too great, and it is difficult to replicate domestic production elsewhere," said vice president Heather Kuang.
Target has moved some orders back to Chinese suppliers, according to two people familiar with the matter, citing supply-chain disruptions and production constraints. Shein is scaling back some Vietnam operations.
None of this is nostalgia. It is triage.
The constraint that actually binds is not price. It is the durability of the legal basis under which a shipment moves. Whatever pretext the president chooses, the US tariff is at present whatever the president wishes it to be — which leaves the courts playing whack-a-mole, treating each claimed authority as sound until it is found otherwise.
No factory amortizes against a rule that can be vacated on a Tuesday.
That is why the rate has stopped functioning as a siting variable. You can build for a 6% tariff. You cannot build for a 6% tariff that a district court, a Court of International Trade panel, or a reissued emergency declaration can convert into 20% or 40% within a product cycle.
The workaround is enormous and unstable. Roughly $300 billion worth of tariffed goods are avoiding the levies annually and reaching the US via Southeast Asia and Mexico, according to shipment-level records analyzed by supply-chain platform Altana. That is not leakage at the margins. That is a parallel trading system.
And it is precisely what enforcement is now aimed at. The US, EU and China have all shifted from writing rules to policing them, which is why legal triage has become a center of corporate due diligence rather than a compliance afterthought.
Under that regime, tariff rates are a forecast, not a fact. The defensible strategy is dual capacity and reversibility: keep the plant you can scale up, keep the plant you can scale down, and keep the legal argument for both alive.
You can see it in the hedges. Yu Yangxian, who sells electric lockers and vending machines, is keeping roughly one-eighth of total capacity in Vietnam. She said the company could expand there again "if Trump goes crazy" and tariffs spike. That is not a supply chain. That is an option.
You can see it in the reliability math. Access to power has become as decisive as price, particularly after the Middle East crisis tested energy reliability across manufacturing bases. Guan Baokui, a Qingdao lawyer advising manufacturers, said Vietnam and Indonesia suffer from "unstable and not continuous" electricity supply, a problem that worsened as oil prices surged.
Stanislaw Krykun, CEO of Poland-based packaging firm DST Pack, sources 80% of production from a Shenzhen factory, with 10% each from backup plants in the US and Europe. Those alternatives cost two to three times more per unit.
He dismissed relocating to Southeast Asia after watching a business partner struggle in Vietnam. "He faced a lot of issues starting with production and finishing with the export," Krykun said. "The system there doesn't really work as smoothly as it works in China." In a crisis, he added, Chinese plants are the most stable you can use.
This is the part tariff models miss. A tariff is a number. A supplier network is a set of relationships, a power grid, a port, a customs broker, a mould shop, and a legal theory.
The legal theory is now the weakest link.
Which is why the flows are not one-directional. Chinese manufacturers are still internationalizing, and their suppliers are following them abroad. Japanese firms are still diversifying portions of production out of China while retaining large-scale operations inside it, per JETRO survey work. Tamura plans to reduce its China production bases; TDK is adding battery-cell production in India while China remains its principal base for those cells.
Both directions expose the same thing: the factory is only part of the calculation.
At the extreme end, SpaceX's reported efforts go beyond removing Chinese production or suppliers and approach supply chains with zero Chinese exposure — scrutinizing supplier personnel, equipment and internal systems. Whether a component is made in China matters less than whether China is embedded anywhere behind it.
That is national security overriding economic geography. It is also, for everyone outside defense, unaffordable.
Malaysian exporters are reading the same map. Invest Penang said automated test equipment companies exporting to the US are actively evaluating final module assembly lines inside the US to circumvent steep tariff increases and hold market access. The Federation of Malaysian Manufacturers, representing 450 US-exporting members, reports members exploring efficiency upgrades, export diversification and contract revisions to absorb renegotiated pricing.
Note what that is: legal positioning, not cost optimization.
Nowhere is the drift clearer than EVs. Chinese vehicles are blocked from the US market by a 100% Section 301 tariff and a Connected Car Rule ban starting in 2027.
Yet Trump recently indicated he might permit Chinese automakers to build factories and produce vehicles in the US. "I am not criticizing Chinese cars themselves," he said, "but the most important thing is that we employ our people."
That is the enforcement logic completing itself. If you cannot stop the goods, you try to stop the geography — and if you cannot stop the geography, you tax the location instead of the origin.
South Korean firms are watching nervously. Hyundai Motor Group sold 1,836,172 vehicles in the US last year against 1,258,730 at home, and its US lineup sits in the mid-to-low price range that Chinese competitors would attack directly. Professor Kim Pil-soo of Daelim University called it a "direct hit" if Washington opens the door after Hyundai's US investment commitments.
The counterargument is that Chinese firms still can't reach price parity here. Imported Chinese battery materials and components carry tariffs, and the IRA's Advanced Manufacturing Production Credit is hard for Chinese-owned plants to capture even when they build on US soil.
Both things can be true. That is the point. The policy is no longer a rate; it is a negotiation with itself.
Meanwhile, the continental picture is softening in one corner and hardening in another. Trump downplayed quitting USMCA and said a deal with Canada could come "fairly soon," after talks collapsed in August and Canada retaliated on Sept 8 with tariffs on US consumer goods and a hike on many US steel products to 50% from 25%.
So the operating environment for a manufacturer in 2026 is this: a rate you cannot trust, a route you cannot fully legalize, a supplier base you cannot replicate, and a court docket you cannot predict.
The firms that survive it are not chasing the lowest tariff. They are buying optionality — two plants, two customs stories, two legal theories — and treating every one of them as reversible.
A 14-point gap favoring Vietnam did not hold production because it was never the binding constraint. The binding constraint was whether anyone could tell you, with confidence, what the rule would be in eighteen months.
Nobody can. So they build where the screws are.