Temu's "Factory-Direct" Prices Weren't a Supply Chain. They Were a Loophole.

A $4 phone case lands on your doorstep from the other side of the planet, and somehow it cost less than the gas to drive to a store that doesn't carry it. Temu told you why: no middleman, no markup, straight from the factory floor to you. It was a beautiful story, and it had a date with a deadline. On May 2, 2025, the United States stopped letting low-value packages from China enter duty-free.⁴ Within weeks, Temu's U.S. weekly sales were down more than 25% from a year earlier - while Walmart, Amazon, and even rival Shein returned to growth.⁶ The factory was still there. The prices were not.

The official story is that Temu cracked the supply chain - cut out the layers, and the savings flow to you. The real story is that most of the magic happened at the border and on a balance sheet you were never allowed to see. Temu's price wasn't an operational edge. It was a regulatory subsidy wearing a supply-chain costume.

The thing it called "factory-direct" was mostly a customs rule

Strip the marketing away and Temu's price advantage rested on two pillars, neither of them a clever logistics trick. The first was de minimis: a long-standing U.S. rule that let any imported parcel under $800 cross the border duty-free. Temu's whole shipping pattern - one cheap item at a time, mailed individually from China to a single buyer - was engineered to slip under that line. Temu and Shein were among the exemption's primary users.⁷ A 'factory-direct' parcel that pays zero duty is not competing on its supply chain. It is competing on a tax holiday that domestic and warehoused importers don't get.

The second pillar was money you couldn't see. PDD Holdings - the parent that owns both Temu and Pinduoduo - does not break out a single line of Temu's financials in any of its SEC filings.¹ What outside analysts can estimate, they don't like. Goldman Sachs pegged the loss at about $6 on every U.S. order; China Merchants Securities estimated Temu was bleeding somewhere between roughly $580 million and $950 million a year. Temu's own spokesperson called those numbers off-base - and then declined to share the real ones.⁵ The low price was real. So was the bill someone else was paying for it.

What a borrowed model can't borrow is profit

Here's the part the origin myth skips. Temu did not invent its supply chain - it inherited one. PDD's Pinduoduo had spent years building a manufacturer-direct model inside China before Temu commenced commercial operations in 2022, plugging into a network already populated by millions of merchants.¹¹⁰ That's why the growth looked effortless: PDD's total revenues jumped 90% in 2023 to roughly $34.9 billion, then climbed to about $54 billion in 2024 as Temu scaled internationally on top of that machine.¹² But those numbers fuse Pinduoduo's China business and Temu together. The headline growth was real; the proof that Temu itself paid for its own prices simply isn't in the filing. A model you can copy off the shelf gets you scale. It does not get you a moat - because everyone shopping at the same factories and shipping under the same exemption has the same prices, and the only differentiator left is who is willing to lose the most per order.

When the loophole closed, the floor fell out

If the price advantage were really operational, killing a customs rule shouldn't dent it much. It demolished it. The U.S. ended duty-free treatment for China on May 2, 2025; parcels under $800 suddenly faced 30% of value or a flat per-item charge.⁴ Temu's U.S. daily active users fell 52% in May versus March, per Sensor Tower data; Temu's U.S. ad spend had already declined 40% year-over-year in April as the company pulled back its customer-acquisition machine in response to the changed economics.⁹ Then came the sales: down more than 25% year-over-year through early June, even as competitors recovered.⁶ By August 29, 2025 the exemption was gone for every country, and the number of sub-$800 parcels entering the U.S. collapsed by 54%.⁷ You cannot watch a price advantage evaporate that fast and still call it structural to the supply chain. It was structural to the regulation.

But isn't every cheap retailer chasing scale first?

The fair objection is that loss-leading to buy a market is an ordinary growth playbook - Amazon did it, Uber did it, and nobody called those arbitrage. There's truth in it: subsidized acquisition can buy habit, and habit can outlast the subsidy. But the comparison flatters Temu. Amazon's losses bought warehouses, logistics, and Prime - durable assets that lowered cost permanently. Temu's lowest prices leaned on a duty exemption it didn't control and an inherited supplier network anyone could also dial. Two of the three things holding the price down - the loophole and the parent's willingness to absorb losses - were not assets Temu was building; they were conditions it was renting. And the moment one of them was repriced by Washington, the engagement it had paid so much to manufacture went with it. Loss-leading builds a moat only if the spending becomes infrastructure. Here, the spending was the moat - and a moat made of subsidy drains the day the pumping stops.

Temu's genius was real, but it was financial and legal long before it was logistical. It found the seam where a customs rule, a subsidized balance sheet, and an inherited factory network overlapped, and it priced into that seam harder than anyone dared. That works beautifully right up until someone notices the seam and stitches it shut. The factory-direct story sold a permanent edge built on a temporary structure - and 2025 is the year the world quietly removed the structure and asked the price to stand on its own. It couldn't. The lesson outlives the discount: a price is only as durable as the rule that allows it.