The Half-Decoupling: Why Moving Out of China Has Not Meant Moving On

Three Desks, One Question

At the head of sourcing's desk in Amsterdam, on a Tuesday morning in early April, is a spreadsheet that should be simple. One brand, two labels, twelve European markets, €25 million of procurement. The columns are country of manufacture, fabric mill, fabric origin, minimum order quantity, lead time, landed cost and a risk rating in four colours. The row causing trouble is a cotton poplin shirt going into SS27, sewn in Turkey, which is where 40 per cent of the procurement book sits. The fabric is from a mill in Bursa, which is where it should be. The yarn inside that fabric is from Xinjiang, which is where it absolutely should not be.

The row is not, technically, her problem yet. The EU Forced Labour Regulation does not fully apply until 14 December 2027, and US Customs and Border Protection has never detained one of her shipments. But it will be her problem the moment it ships to the US, the moment the Dutch competent authority asks for provenance on end-of-life obligations, and already in a narrower, more irritating sense: the Bursa mill will not commit to SS27 fabric without 90 days' notice, because the yarn ships from Shandong, because the spinner there will not commit without 60 days' notice of its own. By the time her buying team decides whether the poplin shirt is a hero piece or a filler, the fabric has already been committed. The brand moved 40 per cent of its production into Turkey so it could respond more quickly to what its European customers were buying. On this row, at this mill, for this fabric, it cannot.

Two time zones west, a product development director in London runs the same conversation through different numbers. Her brand is womenswear, selling direct through stores and online in the UK and the US, shipping into Europe from a UK fulfilment centre, and opening stores in France and Spain next year. Her supplier base is spread across China, India, Vietnam, Cambodia, Bangladesh, Turkey, and Portugal. She has been growing for four years, which has meant adding suppliers, not shifting them. Until now. Her CEO asked last week, after a conversation with a CSRD-reporting retail partner, whether the brand could state its Chinese content by value across the range. She cannot. Nobody in her category can, yet, without commissioning work. The tech packs name the mill. They do not name the mill's suppliers. The question is not whether she has an answer. The question is how much she wants to know.

A third desk, in Miami, belongs to a head of buying at a multi-brand group, running the denim labels. Five years ago, the mix was 70 per cent China and 15 per cent Mexico. Today, those numbers have flipped. She moved deliberately, paid the transition costs, engineered around the yarn-forward rules of origin in USMCA, rebuilt washes, managed the lead-time differentials, and protected gross profit by a margin thinner than she would like. She has been looking at Bangladesh for the better basic denim, at Pakistan for the heavier weights, at a couple of Central American CAFTA-DR options that make sense if the fabric supply base there matures. On 20 February 2026, the US Supreme Court struck down Donald Trump's reciprocal tariffs. On the same day, the administration imposed a flat 10 per cent Section 122 tariff that expires on 24 July 2026 unless Congress acts. Her Mexico costings, built around USMCA duty relief, were fine. Her China costings and her Bangladesh exploration both moved significantly in 48 hours. The board wants to know her AW26 position by early May. She is running two versions of the costing sheet.

Three desks, three jobs, three different expressions of the same underlying question. The question is not the one that trade journalists and politicians have been asking for five years, which is whether the fashion industry is moving out of China. By most published measures, the answer is yes. China's share of US apparel imports has fallen from roughly 40 per cent in the late 2000s to about 11 per cent by the end of 2025. Its share of world clothing exports has dropped to 29.6 per cent in 2024, the lowest since 2010. Bangladesh and Vietnam have taken much of the slack; Turkey, India, Mexico and Central America have taken some. The headline shift is real.

The question growth-stage brands are actually asking, which is much less visible, is different. It is whether the shift they have already made has given them what they moved for. It is whether the fabric flowing into their non-China suppliers is letting them run faster, or forcing them to commit earlier. It is whether the MOQs in their new countries are letting them breathe. It is whether, when a style sells, they can chase it. It is whether, when a style dies, they can stop. It is whether they have built a supply chain that serves the business, or a supply chain that the business now serves. And it is whether the regulatory perimeter closing around them over the next 24 months will reward or punish the decisions they have already taken.

This report is about that question. The statistical shift out of China is real, but it is thinner than it looks, because Chinese fabric, Chinese capital and Chinese ownership sit upstream of most of the non-China production that Western brands have been buying. That thinness matters less as a political problem than as an operational one. It is why a brand that has moved 85 per cent of its assembly out of China may still find that 30 to 45 per cent of the value of a Vietnamese garment, by origin, is Chinese. It is why a Dutch menswear director with 40 per cent in Turkey cannot activate Morocco without first solving the fabric supply. It is why the Prato workshops in Tuscany, producing garments with "Made in Italy" labels, are running Chinese fabric through Chinese-owned factories with Chinese workers, and why nobody in European fashion quite knows what to do about it. And it is why fluency, not decoupling, is the honest frame for what sourcing teams are trying to achieve in 2026 and 2027.

How Fashion Sourcing Actually Shifted

What has happened to apparel sourcing over the past decade looks, from a distance, like a straightforward reallocation. Look closer and it divides into two quite different stories: the story of where the sewing machines are, and the story of where the value comes from. The first story is the one the trade statistics tell. The second story, until recently, has been much harder to see.

The first story is clear enough. According to the World Trade Organisation's statistical review for 2024, China's market share of world clothing exports fell to 29.6 per cent, the lowest level since 2010, down from roughly 36 to 37 per cent around 2010. Bangladesh has held the number two position, exporting around $39 billion in the fiscal year that ended in mid-2025 with a global share of just under 8 per cent. Vietnam sits at number three with roughly $31 billion and a 6 per cent share. Turkey, India, Indonesia, Cambodia and Pakistan round out the top ten. In the United States, the pattern is sharper still. Analysis of Office of Textiles and Apparel data shows China's share of US apparel imports has collapsed from roughly 40 per cent in the late 2000s to around 11 per cent in late 2025, with import values from China down more than 50 per cent year-on-year at some points. US fashion companies now routinely source from ten or more countries. The USFIA's 2025 benchmarking survey found that 100 per cent of respondents source from Vietnam, 94 per cent from Cambodia, 88 per cent from Bangladesh, 77 per cent from Indonesia and 53 per cent from Sri Lanka, up sharply on the year before.

The brand-level picture tracks the aggregate. In recent earnings calls, Adidas has stated that China is "almost irrelevant" for its US business, with Chinese imports reduced to 2 per cent and Chinese capacity now reserved for China-for-China supply. Puma has emphasised Vietnam, Cambodia and Indonesia. Carter's has named Vietnam, Cambodia, Bangladesh and India as its largest countries of origin. Abercrombie and Fitch expects Chinese US sourcing to stay in the low single digits. Gap Inc. disclosed in early 2025 that less than 10 per cent of its production is from China; in October 2024 the company sourced from just 36 Chinese factories, representing 7.2 per cent of its apparel base. The direction is unambiguous.

So is the reason. It is not, principally, a preference story. It is a risk and arithmetic story. The Trump administration's first use of tariffs against China in 2018 and 2019 began the movement. The pandemic accelerated it. The Uyghur Forced Labour Prevention Act, in force since June 2022, made specific categories of Chinese apparel exports commercially unviable into the US. Then, in April 2025, the administration's so-called "Liberation Day" tariffs lifted the average US apparel duty from roughly 14 per cent to 23.8 per cent by May 2025, with reciprocal rates as high as 49 per cent for Cambodia, 46 per cent for Vietnam, 37 per cent for Bangladesh and 32 per cent for Indonesia. These were reduced in bilateral deals through the summer, and then largely struck down by the Supreme Court on 20 February 2026 in a 6-3 ruling against the use of the International Emergency Economic Powers Act. The administration responded the same day with a flat 10 per cent tariff under Section 122 of the Trade Act of 1974, which carries a statutory 150-day limit and expires on 24 July 2026 unless Congress acts. Section 301 duties on China, applied on top, remain in force. Section 232 sectoral tariffs stack. Refund claims for IEEPA duties paid between April 2025 and February 2026 are now in play. The planning horizon for a US-facing brand has collapsed to roughly 90 days.

Europe has had its own version of this. The EU remains the world's largest apparel importer and, since 2024, has slightly overtaken China to become the largest apparel exporter by value, at 29.7 per cent of global share against China's 29.6 per cent, driven mainly by intra-EU trade and a rebuilding of proximity production. European brands have leaned more heavily on Turkey and North Africa as nearshore partners, though as will become apparent, that story has unravelled during 2025. The European regulatory story is different in shape. It runs through the Corporate Sustainability Due Diligence Directive, the Corporate Sustainability Reporting Directive, the Ecodesign for Sustainable Products Regulation and its Digital Product Passport, the Waste Framework Directive with its EPR provisions, and the EU Forced Labour Regulation. The Commission's Omnibus I simplification package, adopted by the Council on 24 February 2026, has narrowed the scope of CSDDD and CSRD to companies above 5,000 employees and €1.5 billion in turnover, and above 1,000 employees and €450 million respectively. In a narrow reading, this lifts the direct regulatory burden from almost every brand in the growth-stage range. In practice it does nothing of the kind: compliance obligations flow downstream to the supplier base in the form of data requests, and growth-stage brands supplying CSDDD-scoped companies, or sharing their customers, will have to meet the same standards via the market rather than directly via the law.

So far, so familiar. This is the landscape as the trade press has described it. The second story, which matters more to how a brand actually runs its supply chain, is the one that headline trade data obscures. It concerns the origin of value, not the origin of assembly.