Can Amazon keep up with Temu, AliExpress, and the Chinese price?
Nathan Chadwick · · 17 min read
Listen
TLDRRead the short version
- The price on Temu, Shein and AliExpress is a stack: a real factory advantage, fewer middlemen, a subsidised first order, and a duty-free parcel. It is a different product from a Prime delivery.
- Alibaba.com is the wholesale layer. AliExpress is the consumer parcel, now being pulled towards local stock. Temu is PDD setting the price. Shein is a fashion factory that cuts another batch only when the last one sells.
- The duty-free parcel is closing. The US ended it for China and Hong Kong on 2 May 2025 and for every country on 29 August 2025. Since 1 July 2026 the EU has charged a temporary €3 duty on low-value items, and since 1 November 2026 a €2 handling fee on top.
- Amazon already sells that factory. Chinese sellers are a majority of the names on Amazon and a minority of the money at the top. The fee stack, the van, and the return are why the same cable costs more.
- Haul is Amazon's slow aisle. In March 2026 it was still about $2 billion of US sales against a Temu business in the tens of billions. By November 2026 Amazon was cross-listing ordinary stock into Haul and showing it in normal search.
On this page
The price and the promise are different products#
People keep asking whether Amazon can match Temu. The question treats them as two shops with the same job. They have the same factories in the background and a different promise on the screen.
One is a search box that says the parcel can be here tomorrow, and that sending it back will be boring. The other built its name on a price that looks wrong and a wait that used to be measured in weeks. A growing share of those orders now ships from a warehouse in the destination country, some of them in a couple of days. AliExpress sits between them. Alibaba.com sits behind both, selling the carton to whoever is about to become an importer. Shein does a narrower version of the trick in clothes.
I think Amazon can keep up with the cost of the goods. That cost is already on Amazon, listed by the people who make the goods or by the trading companies one step away. The price people screenshot from Temu in 2024 was that cost, minus the middlemen, minus the duty, minus whatever PDD was willing to spend so you would install the app. Amazon can meet the factory. It meets a Prime promise by charging for the rest. Since the middle of 2025, the duty piece of that screenshot has been going away.
Four Chinese businesses, four prices#
Alibaba the company is a group. The site a small retailer means by "Alibaba" is usually Alibaba.com, the international wholesale marketplace. You buy in cartons. Freight, duty, a warehouse, returns, an advert, and a wage all come after the number on the page. Compare that number with a price on Amazon and you are comparing a supply price with a finished offer. A lot of Amazon sellers, and a lot of Shopify shops, start exactly there. The Chinese price people think Amazon is failing to match is often the price Amazon's own sellers pay before they list.
AliExpress is the consumer end of the same pipe. You buy from a seller, a parcel turns up, and for years the deal was simple: wait, pay less, accept that some of it will be rough. The part of AliExpress that matters now is Choice, where the platform takes more control of the price, the logistics, and the service. In the year to 31 March 2026 Alibaba said Choice's unit economics kept improving, and it launched AliExpressDirect on stock already sitting in more than 30 countries, plus a Brand+ programme for labels going overseas. AliExpress covers more than 200 countries. The long wait for a 90p cable is still in the app. The company is wrapping it in local stock and brands, because the naked version got expensive to run and easy to regulate.
Temu is the one that moved the anchor. It belongs to PDD Holdings, which also runs Pinduoduo in China. The original overseas model was fully managed. A factory offers a supply price, Temu sets what you pay, the goods are consolidated, and a parcel leaves. The factory does not own the customer. For a while the price on the screen was a customer-acquisition price: shipping and advertising plugged by the rest of PDD. The shift off the pure China parcel started before the loophole closed. Marketplace Pulse reported that in July 2024, 20% of Temu's US sales already came from local warehouses, up from none at the start of that year, and that Temu was aiming for half of UK sales to be local-to-local by the end of 2025 and for 80% of European sales to move through local warehouses. Some of those orders were already arriving in two days. When the US shut the small-parcel exemption, that semi-managed model stopped being an experiment. Sellers hold stock closer to the buyer, Temu still leans on the price, and the operation starts to look like a marketplace with warehouses. That is closer to the game Amazon already plays.
Shein gets lumped in with them, and the lump is only half right. Shein is closer to a retailer that owns the design and the reorder. A style goes up, a small batch gets made, and another batch is cut if it sells. The advantage is the speed from a factory district to a product page. A general marketplace does not copy that loop by opening a bargain tab. I do not have Shein's private accounts. The public lesson is enough: in apparel, the cost people feel is a production system, and the parcel loophole was an accelerator.
TikTok Shop is the other name that belongs in the "etc." It sells through a video, and by 2025 it had reached a US gross-merchandise band that Marketplace Pulse put near Walmart's marketplace, around $15 billion. Discovery is the product. Price still has to clear. It is a third way of finding the same factories.
Where the attractive cost actually comes from#
"China is cheaper" is true at the factory gate and useless as an explanation of the app.
The factory advantage is old. Clusters that make electronics, homewares, toys, and clothing have the tooling, the suppliers, and the labour density next to each other. A simple product designed to a price, made next door to the mould, leaves the gate cheaper than the same sketch would in Britain or the US. Amazon's shelves have been full of that for a decade. Temu did not invent Dongguan.
The second layer is the missing cut. A brand, a wholesaler, an importer, and a shop each used to take something. AliExpress and Temu collapse that into a platform and a parcel. Alibaba.com hands the importer's job to the buyer. A £12 product in a British shop against a £4 product in an app is, in large part, those cuts. The object can be the same object.
The third layer was the rule, and it is the layer that has moved.
In the US, a parcel worth $800 or under could enter without duty. Temu and Shein industrialised that. On 2 May 2025 the exemption stopped for China and Hong Kong. On 29 August 2025 it stopped for every country. US Customs and Border Protection has since said it collected over $1 billion in duties on more than 246 million of those low-value shipments, and that seizures of unsafe and non-compliant goods in the stream rose 82%. The White House was explicit that China and Hong Kong had been the majority of the parcels.
Europe had its own version: no customs duty under €150. The Council, citing the Commission, puts the 2024 flow at 4.6 billion such parcels, 91% of them from China. The Commission later put 2025 at almost 5.9 billion low-value items shipped straight to consumers. From 1 July 2026 a temporary €3 duty applies per item category in consignments up to €150, until normal tariffs take over in 2028. From 1 November 2026 a €2 Union handling fee applies per item category on distance sales from outside the EU, including on parcels above that value. Ireland's Revenue Commissioners describe the fee as aimed at the platform, and non-refundable when the goods come back. On a single-category parcel under €150 that is €5 before VAT. On a product that was £3, the rule is now the same size as the product.
Britain met this earlier, on tax. Since 1 January 2021, consignments of £135 or less sold through an online marketplace have VAT collected at the checkout, usually by the marketplace. A Temu order in the UK has shown tax for years. The American and European changes are the duty version of a hole that, on VAT, was already shut here.
The fourth layer is the subsidy. PDD does not publish Temu on its own, so anyone who quotes a Temu profit is guessing. What PDD does publish is a group that still makes a lot of money, grows more slowly, and spends a fortune on promotion. In 2025, revenue was RMB431.8 billion, about $62 billion, up 10%. Operating profit fell 13% to RMB94.6 billion. Net income fell 12% to RMB99.4 billion, about $14.2 billion. Sales and marketing was RMB125.3 billion, about $18 billion, up 13%. In the quarter to 30 June 2026, revenue rose 8% to RMB112.4 billion and net income fell 12% to RMB27.2 billion. Management said the investments would affect the numbers. That is the polite version of "the price on the screen is partly a budget."
The week the US exemption went, the budget met a rule. Sensor Tower figures reported by CNBC had Temu's US daily users down 52% in May 2025 against March, and Shein's down 25%. Reuters reported the same near-halving. A price that needs a loophole and an ad slot moves when either one is pulled. PDD's China business can carry a lot. It is already being asked to.
Alibaba's international accounts are the calmer version. In the year to 31 March 2026, Alibaba International Digital Commerce (AliExpress, Lazada, Trendyol, Alibaba.com) did RMB144.2 billion of revenue, about $21 billion, up 9%. The adjusted EBITA loss narrowed from RMB15.1 billion to RMB2.1 billion. In the March quarter the loss was RMB138 million, which Alibaba called approaching break-even. This is a cross-border retailer that spent years buying the customer and is now being run so the parcel pays for itself.
Amazon already has the factory. It charges for the rest.#
The scare is that the apps take Amazon's customers because they are cheaper. The seller file says the factory moved in years ago.
In September 2025 Marketplace Pulse put Chinese sellers at 50% of Amazon's active sellers worldwide, the first time they had crossed that line across the marketplaces. On its estimate, US sellers still took about $157 billion of Amazon.com's third-party sales against about $132 billion for Chinese sellers. By July 2026 Chinese sellers were 55.9% of the top 10,000 sellers on Amazon.com, up from 42.5% in July 2020. American sellers in that list had fallen from 53.7% to 40.5%.
Headcount is the wrong scoreboard. In that same July cut, US sellers produced 65.3% of the cohort's sales and Chinese sellers 28.6%. In the top 100, American sellers were 81.4% of the names and 93.2% of the money. Their average selling price there was $47.62, against $22.03.
That pair is the essay. China holds the catalogue of cheap, replaceable goods. The brands and the dearer baskets still throw off most of the money, and a large share of those names are American. Amazon takes a fee on both. The cable costs more on Amazon than on Temu because referral fees, fulfilment, storage, returns, and advertising sit on the factory price. Marketplace Pulse has had fees as the first thing sellers complain about, with advertising close behind. The same factory can list in both places. Only one of those places includes a van this week and a returns desk.
Amazon's quarter to 30 September 2026 shows what that machine earns. Net sales were $201.9 billion, up 12%. Prime Day had moved into the previous quarter; Amazon said growth would have been about four percentage points higher without the shift. Inside the quarter, product sales were $75.3 billion and service sales were $126.6 billion. Third-party seller services were $46.5 billion, up 9%. Advertising was $19.6 billion, up 11%. The first-party online store grew 2% to $68.5 billion.
Operating income was $28.1 billion. AWS made $17.8 billion of it, on $46.0 billion of sales, growing 39%. North America, the store plus the ads and Prime counted in that segment, made $8.7 billion on $112.9 billion. That is a bit under 8%. International made $1.6 billion on $43.1 billion. On AWS, operating income was just under 39 cents per dollar of sales. The shop is a real business with a shop's margin. The cloud is a different business, and it is large enough to fund a bargain aisle while the shop is told to keep its promise.
Andy Jassy's frame in that quarter is the right size for the panic. Nearly 80% of retail is still physical. Amazon is trying to take the next chunk of that with speed, groceries, and a lower cost to serve. The Chinese apps are fierce in the slice of the basket that can wait, and that slice is large. It is a slice.
Haul is the aisle, and search has started to blur it#
Amazon launched Haul on 13 November 2024. Items under $20, a lot of them from China, delivery in something like a week or two. The point was to keep the cheap errand inside the app you already trust.
By 5 March 2026, Marketplace Pulse counted 3,287 US Haul sellers and about $2 billion of annual gross merchandise, built from seller economics. In that article it put Temu near $30 billion of US gross merchandise and Shein near $18 billion. Two weeks later, in a wider marketplace ranking, it put Temu's 2025 US business nearer $22 billion, with TikTok Shop and Walmart's marketplace around $15 billion each. Those are estimates, and they do not agree to the dollar. The order of magnitude does. Haul was a real aisle. Temu was an order of magnitude larger in the US, and Amazon's own third-party business is an order of magnitude larger again. Marketplace Pulse has put Amazon.com third-party sales around $300 billion.
Amazon's own update, on 9 November 2026, talks about selection and shoppers. More than 10 million products, a catalogue more than 25 times larger than two years earlier, purchasing customers up nearly 350% year over year, most items under $10, more than a million under $5, more than half a million under $3. Free delivery from $25, 5% off over $50, 10% off over $75, and free returns on eligible items over $3 within 15 days. The 350% has no base published under it. A small shop can triple and stay small. Read it as evidence that people will use the aisle.
The same week, Amazon said relevant Haul products now appear in ordinary search. On 12 November, Marketplace Pulse counted nearly 184,000 sellers with a listing in Haul, about four in ten active Amazon.com sellers. Most of them did not apply. Amazon cross-listed them, including stock already in its warehouses, through Haul Distributed Fulfillment. The original direct-from-China route is still invitation-only, which is why the jump from 3,287 sellers to 184,000 is mostly reclassification. A DeWalt bit set was in Haul at $9.98, sold by Amazon, with next-day Prime delivery. Sellers pay the normal fees. Amazon funds the Haul promotion. Leaving means taking the whole account out.
That is a different strategy from "build a Temu." A separate slow store keeps Prime as Prime. A price band stamped onto the existing catalogue, then dropped into the same search results, defends the feeling that Amazon has the cheap version. It also puts a $4 lookalike next to a $14 product that was never a cross-border gamble. For a brand in home, beauty, or toys, the competitor in the results can now be Amazon's own tag.
Outside the US the same idea has been running as Amazon Bazaar. By November 2025 Marketplace Pulse had Haul and Bazaar across more than 25 markets, including the UK, Germany, France, Italy, and Spain, and in March 2026 it still described Bazaar as a global expansion on that scale. The slow aisle is a response to a shopping habit that was never only American.
Can Amazon keep up?#
With the cost of making the object, yes. The factory price is on the marketplace, and the seller data says the cheap end of the catalogue has already been won by the people closest to that factory. With the fully loaded price of a two-day parcel and an easy return, Amazon keeps up by being the more expensive, more reliable version, and the accounts say customers still pay for that on the baskets that matter.
With the 2024 Temu screenshot, a $4 thing on a duty-free slow parcel with the first order subsidised, Amazon should leave it alone inside Prime. Matching it there means burning the promise that pays the North American margin. The rule that made the screenshot so extreme has been withdrawn in the US and taxed in the EU. Temu's warehouses are the tell that the next fight is fulfilment. A fifth of its US sales were already local by the middle of 2024, and a slice of orders now moves in days. That is the part of the stack Amazon already owns, and it is a fight with ordinary margins, fought by PDD's marketing line as well as by Amazon's vans.
What Amazon has to keep up with is the habit. For fashion, home bits, and anything you can wait for, a lot of people no longer start in the Amazon search box. Haul is the attempt to be in that moment and leave the rest of the shop alone. The November search change is where that attempt can undo itself. Cross-listing a Prime-speed branded set into the bargain aisle trains people to hunt the tag on goods that were supposed to stay in the main store.
The money lets Amazon be patient. AWS and advertising can carry a bargain aisle while it finds a size. They will have a harder time carrying a shop in which every replenishment basket gets repriced to a screenshot. The attractive Chinese cost is a factory, a thinner chain, a customs treatment that is being withdrawn, and a budget PDD is already telling investors will show up in the profit. Amazon can meet the first two. It can offer the slow version of the third as Haul. The budget was the part that made the apps look unbeatable, and budgets are a choice.
What would make this wrong#
Haul becomes the store. If customers learn to wait for the Haul price on goods that used to sell at the normal one, Amazon will have cut its own fee to answer a competitor that was never going to win the weekly shop. Putting Haul into the main search results is how that happens without a strategy memo.
Local stock gets fast enough. Some Temu orders already arrive in a couple of days. If that becomes the normal parcel, at a price the Amazon fee stack cannot touch, speed stops being a defence. Amazon is built for that fight. The fight is open, and it is further along than a 2023 memory of Temu suggests.
The duties fail in practice. A €3 charge and a closed American exemption matter only if they are collected. Undervaluation, split parcels, and a change of government can hand the gap back. The 2024 price was a policy outcome. Policy outcomes move.
The compliance gap was the cost gap. CBP's seizure figures are the dull version of the bargain. Counterfeits, unsafe electronics, and goods that miss the rules are one way a catalogue stays cheap. If platforms are held to the product, a slice of the gap was a compliance gap. Amazon's A-to-z guarantee and its returns are why plenty of people pay the extra, and they are a cost the bargain apps are being pushed towards.
The fee stack drives the good sellers off. Referral fees, fulfilment, and ads can leave a careful seller with nothing, and then the factory simply lives on Temu and AliExpress. The July 2026 figures say the money has stayed. That can change, and it would change the argument faster than another million Haul SKUs.
The comparison people want is Amazon against a Chinese price, as if one of them is about to lose. Amazon can keep up with the cost. It does it by selling the same factory through other people's accounts, and by running a separate slow aisle for the basket that was only ever about the price. A Prime parcel stays a different product. After this year, the parcel that made the apps look untouchable has a customs form attached.
References#
- PDD Holdings, fourth quarter and fiscal year 2025 results (25 March 2026)
- PDD Holdings, second quarter 2026 results
- Reuters, Temu-owner PDD misses quarterly revenue estimates (24 August 2026)
- Alibaba Group, March quarter and fiscal year 2026 results (year ended 31 March 2026)
- Amazon, third quarter 2026 results
- Amazon Form 10-Q for the quarter ended 30 September 2026
- Amazon, Andy Jassy on stores growth, third quarter 2026
- Amazon, Haul early holiday sale, selection and search (9 November 2026)
- Marketplace Pulse, Amazon Haul launches (13 November 2024)
- Marketplace Pulse, the supply-chain reversal: Temu's local warehouses and Amazon Bazaar (20 November 2025)
- Marketplace Pulse, Haul passes 3,000 sellers (5 March 2026)
- Marketplace Pulse, top 10 US marketplaces in 2026 (19 March 2026)
- Marketplace Pulse, four in ten Amazon sellers are on Haul (12 November 2026)
- Marketplace Pulse, Chinese sellers reach a global majority on Amazon (25 September 2025)
- Marketplace Pulse, Chinese sellers in Amazon.com's top 10,000 (9 July 2026)
- CBP, end of de minimis enforced worldwide (29 August 2025)
- CBP, more than $1 billion collected on low-value shipments since the phase-out
- The White House, de minimis suspended worldwide from 29 August 2025
- CNBC, Temu and Shein US demand after the loophole closed (5 June 2025)
- Reuters, Temu US daily users nearly halve (2 June 2025)
- European Council, EU action on small parcels, including the €3 duty and the €2 handling fee
- European Commission, temporary €3 duty on low-value imports from 1 July 2026
- Commission Delegated Regulation (EU) 2026/2152, Union handling fee of €2
- Irish Revenue, Union handling fee from 1 November 2026
- HMRC, VAT on overseas goods sold to UK customers through online marketplaces