What is Pinduoduo betting on by not chasing AI hype?

**Key points**

**1. Pinduoduo is not avoiding AI; it is publicly silent about it**

In 2025, reports said it had formed a large-model team and was hiring AI talent intensively, yet it dodged AI on every quarterly earnings call; R&D spending has long been only about 3% of revenue.

**2. Its moat is demand aggregation, not low prices**

The platform aggregates scattered demand into confirmed orders that factories dare to bet on; the algorithm allocates traffic by price and conversion, letting merchants trade profit for traffic while the platform holds pricing power.

**3. A low monetization rate becomes a weapon in price wars**

**4. For Pinduoduo, AI is only a matter of efficiency**

**5. It is betting that cheap fulfillment remains the deciding factor; the risk is AI reshaping shopping**

It is betting that the next decade will still be about delivering goods to people's hands more cheaply, and no company can yet use AI to turn 9.9 yuan into 8.9 yuan; Temu's losses are narrowing. If AI assistants reshape the shopping path, the demand-aggregation advantage could fail, and the money it saved could have bought it a transition ticket. 2026-11-02 10:39 Why isn't Pinduoduo chasing the AI hype, and what is it betting on? Jiexin Quick Take

This article is from the WeChat public account: Jiexin, author: Jiexin Jiege Xin

As early as 2025, reports mentioned that it had already formed an internal large-model team, led by a core member from Baidu's Phoenix Nest, with a salary several times what the person earned at Baidu. Yicai also reported that within a month it was intensively recruiting talent in AI and large models.

Alibaba talks about AI, JD talks about AI, Meituan talks about AI, and ByteDance treats AI as the company's main line. Only Pinduoduo, quarter after quarter, dodges the question. R&D spending has long been only about 3% of revenue, which is almost zero compared with Alibaba's publicly pledged investment of hundreds of billions.

In the second quarter of 2026, Pinduoduo's revenue was 112.4 billion yuan, up 8% year on year; net profit attributable to shareholders was 27.2 billion yuan, down 12% year on year. Revenue is rising and profit is falling, and this is the result it actively chose. In the same quarter, R&D spending rose about 40% year on year, and the company's explanation was "trading profit for ecosystem."

But even after rising 40%, its R&D spending as a share of revenue remains in the lowest tier among big internet companies. Meanwhile, its cash and short-term investments on the books have accumulated to the scale of hundreds of billions.

A company with hundreds of billions in cash and profits ahead of its peers explicitly choosing not to put money into the hottest track cannot possibly be because it did not see it. It saw it and chose not to do it.

In other words: this is a deliberate choice made by a commercial company after careful consideration.

This brings us to Pinduoduo's moat.

2. Its real moat is not cheapness

Many people think Pinduoduo's barrier is low prices. Low prices are not a barrier; low prices are a result.

If a company can sell cheaper than others over the long term and still make money, it must have built a structure in some link that others cannot replicate. For Pinduoduo, that link is demand aggregation.

Think about how a 9.9-yuan free-shipping pack of tissues comes about. It is not because the factory is merciful; it is because on this platform ten million people want to buy the same kind of tissue at the same time, so that demand is compressed into a predictable large order. Only with that order does the factory dare to stock materials all at once, dare to open a dedicated production line, and dare to push costs to the extreme.

What Pinduoduo really does is aggregate countless scattered, tiny demands that originally could not be gathered together into a confirmed order that factories can bet on. That is its core capability.

And what supports this aggregation is its traffic distribution mechanism.

Other companies' recommendation algorithms are essentially about letting merchants spend money to buy exposure, pay per click, and whoever bids higher ranks first. Under this business model, the platform earns advertising fees, but the downside is the Matthew effect in which the strong get stronger.

Pinduoduo's algorithm has a much simpler goal: price and conversion. If a product wants traffic, the most effective way is not to increase the budget but to cut the price. This is why the snowball can roll.

This difference looks like mere product design, but it is actually a difference in power relations. Elsewhere, merchants use money to buy traffic; here, merchants use profit to trade for traffic. Once this mechanism runs smoothly, the platform permanently gains the initiative over pricing.

The best validation of this mechanism was the later 10 Billion Subsidy program. Many people thought it was just a promotion, but in fact it did something harsher: it proved that Pinduoduo has the ability to push down the prices of branded goods as well. When a platform can make white-label factories bid according to its rules and also make brands accept the same rules, its bargaining power across the entire supply chain truly takes shape.

3. Turning "low gross margin" into a weapon

This moat has an even harsher version.

Pinduoduo's monetization rate is the lowest. This looks like a weakness, but in competition it is a nuclear weapon.

Because any rival that wants to fight it in a price war must first do one thing: give up its own monetization rate. And giving up the monetization rate means cutting those rigid expenditures. JD's logistics costs are rigid and cannot be cut; Alibaba's cloud and AI investment is a story the capital market watches and cannot be stopped. Only Pinduoduo never intended to support these things from the start.

This looks like a victory of business model, but it is actually a victory of cost structure. It has pushed per-capita efficiency to the highest in the industry and compressed unnecessary spending to the lowest, so it can stay in the low-gross-margin range indefinitely, while rivals cannot.

Alibaba has its cloud, and AI can directly turn into computing revenue on the cloud. JD has a massive logistics network and self-operated system, and AI can directly lower fulfillment costs. Tencent has social, ByteDance has content distribution, and AI is the fuel for their core businesses. For these companies, doing AI means "investment can turn into revenue or defend the core base."

Alibaba has publicly listed AI as a core investment direction for the coming years, and the investment scale rumored in the outside world is on the order of 300 billion. JD's investment in large models and supply chain algorithms has also never stopped. For them, the money spent buys non-replaceability, not a little more profit.

Matters of life and death must ignore cost; matters of efficiency can proceed slowly.

V. What is Pinduoduo betting on?

This brings us to Pinduoduo's judgment after deep deliberation, or rather, its bet.

This judgment may not be entirely correct, but it has a very solid support: so far, no company has been able to use AI to turn a 9.9-yuan item into an 8.9-yuan one. AI optimizes information and decision-making, not physical handling. As long as goods still have to leave the factory, go through trunk lines, enter sorting centers, and finally be delivered to your doorstep by a rider, the costs on this chain cannot be erased by large models.

And on this chain, Pinduoduo is already far ahead.

So its silence is not sluggishness; it is a very expensive form of self-discipline.

There is also a piece of circumstantial evidence that is easy to overlook: Temu. In the years since it went overseas, it has used almost the same set of things, replicating the domestic playbook of demand aggregation plus extreme cost control into a market with more dispersed supply and longer channels. It lost money overseas for a long time, but judging by the financial reporting caliber, losses are continuously narrowing and the unit economics are improving.

This shows that Pinduoduo's set of capabilities is not something that can only survive in China's soil. It is more like an organizational capability than a local dividend.

What if AI really reconstructs the act of "buying things"? If people in the future no longer "browse" platforms, but directly tell an assistant, "Help me restock everything the house is short of," then demand aggregation will no longer be the platform's job. At that point, Pinduoduo's moat will turn into a dried-up riverbed overnight.

So its choice is essentially to stake all its bets on "change won't come that fast."

This bet has won over the past decade, and won beautifully. But it also means that if the bet is wrong, it won't even have a card to play to turn around. Because every cent it saved could originally have been used to buy that card.

The hardest thing for a company is not deciding what to do, but deciding not to do something when everyone else is doing it. Pinduoduo has achieved this, and that in itself is a capability. But a capability used in the right place is called foresight, and used in the wrong place is called stubbornness; the difference between the two often takes many years to become clear. AI industry signal channel: Frontier Technology. This content comes from the internet, original link; the views represent only the author and do not represent Huxiu's position. For copyright issues, please contact hezuo@huxiu.com, and we will verify and handle it promptly. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]