Double 11 Enters a Protracted Battle: Have the Big Tech Giants Made Peace with AI?

By Lu Jiu Business Review

More than six hundred years ago, Zhu Yuanzhang relied on "build high walls, stock up grain, and delay claiming kingship" to wait for his opportunity in an era when many contenders rose at once. Today's e-commerce platforms certainly do not need to copy the ancients' military strategy, but in an era of slowing traffic growth, fierce price competition, and increasingly rational consumers, these nine characters still carry real-world meaning.

Two years ago, the story about AI was told on a grand scale. Some said the search box would soon become history; some judged that consumers would no longer need to open Taobao, JD.com, or Douyin—they would simply say to a dialog box, "Help me buy a washing machine," and AI would search for products, read reviews, compare prices, claim coupons, and finally place the order directly.

If that day truly arrived, then the business logic that e-commerce platforms spent more than a decade painstakingly building—search advertising, bidding rank, traffic distribution, merchant operations—would seem to be uprooted entirely. Whoever controlled the AI entry point could intercept users; whoever first turned AI into a shopping entry point might reshuffle the deck.

But when the major promotion actually began, things did not unfold according to the script. The imagined platform war did not break out, and AI did not suddenly kick open the door to the old world. It was more like a new colleague, assigned by each platform to different departments: some went to help users make choices, some went into warehouses to calculate inventory, some went into content venues to improve conversion, and some simply walked into factories to study how the next batch of goods should actually be produced.

This does not look like a revolution; it looks more like a redivision of labor.

Lu Jiu Business Review believes that the most noteworthy thing about this year's Double Eleven is not how dazzling the AI features released by any platform are, but that the big companies' attitude toward AI has changed: from "who will subvert whom" in the past two years to "who can make good use of whom" today. This reconciliation has at least three layers of meaning: acknowledging AI's capability boundaries, seeing one's own survival genes clearly again, and beginning to recalculate the cost of competition.

If we borrow the advice Zhu Sheng gave Zhu Yuanzhang more than six hundred years ago to summarize it, it is: build high walls, stock up grain, and delay claiming kingship.

01 Build high walls: AI did not tear down the walls; it entered within them

What Taobao and Tmall most need to solve is the problem that consumers find it increasingly difficult to make decisions. Buying a washing machine looks like an ordinary purchase, but before actually placing an order one faces a series of troubles: how to choose capacity, whether a drying function is necessary, which brand is more durable, which comments are actual experiences and which are just marketing talk, and which model is truly cheaper after stacking discounts. Information is not scarce; what is truly scarce is the ability to organize information into an answer.

Therefore, Taobao AI is more like a shopping decision assistant. Users explain their budget, family situation, and usage needs, and AI helps filter products, compare parameters, summarize reviews, and then sort out discount benefits clearly. For merchants, AI can also participate in hit-product forecasting, product diagnosis, and price management. What it needs to do is not turn Tmall into another chat app, but make the original transactions easier to happen and make brand users hesitate a little less.

JD.com's focus is different. For a small item, consumers may care more about price; for refrigerators, air conditioners, and washing machines, what truly affects the experience is often whether the product can be delivered, when it will be delivered, whether the old machine can be collected, and whether installation is timely. When major promotion orders surge, inventory, warehousing and distribution, delivery, and trade-ins become a complex collaborative exam.

These things may not be suitable for display in front of consumers, yet they may directly determine the success or failure of a transaction. If AI can judge in advance which regions will see rising demand, help warehouses transfer inventory, arrange delivery capacity, and connect old-machine recycling, it can make JD.com's supply chain advantages more solid. It may not make users feel "really smart," but it may make users feel "really worry-free." For JD.com, that is value.

English text is being prepared.

AI can help merchants generate and screen content materials, and can also identify signals of users shifting from "just browsing" to "seriously comparing," connecting the content venue with transaction scenarios such as the mall and search. What Douyin wants to do is let users not only place impulsive orders because of a video, but also come back to find products when they have clear needs. In the end, what it needs to repair is the easily broken path from planting desire to completing a sale.

Pinduoduo's story may be farthest from the consumer's screen, yet closest to the source of business. For industrial-belt factories, the biggest headache may not be the inability to produce, but not knowing what to produce or how much to produce. If goods are made and do not sell, inventory piles up in the warehouse; if they wait to see a product sell well and then follow up, they may miss the window.

If the transaction data accumulated by the platform can help factories test models, judge demand, and arrange production scheduling, there is an opportunity to reduce trial and error and overstock. Pinduoduo's low-price competition appears on the surface to be price competition, but at a deeper level it is competition in supply chain efficiency. The place where AI can truly play a role is not just pushing prices on the page a little lower, but reducing waste among production, inventory, and demand.

What Xiaohongshu protects is trust. When users go there to search for a skincare product, a restaurant, or a camera, they often are not just looking to see who sells it cheapest; they also want to know "how it actually works out for ordinary people." As commercial content increases and genuine experiences mix with carefully packaged ads, the trust the platform depends on for survival is eroded bit by bit.

AI can help identify the semantics of content, determine the type of a post, and assist the platform in spotting watered-down content and unnatural marketing expressions. But there is also a boundary here: algorithms can assist identification, but they cannot automatically equal truth. What Xiaohongshu truly needs to protect is not the myth that "AI can identify everything," but rather that users remain willing to believe the experience sharing here has reference value.

In the past, some believed that once technology arrived, the accumulated assets of old platforms would instantly reset to zero. That judgment overlooked one fact: technology itself does not automatically constitute a business model. A platform's supply chain, merchant network, content ecosystem, user habits, and service capabilities are the "body" formed through long-term operation; AI is more like a newly grown hand. The hand can be more dexterous, but it must obey the body's commands and also know how far it can reach.

02 Accumulate grain: traffic is getting more expensive, so first do the math on your own business

Looking back at past Double 11s, the action platforms were best at was attack: threshold discounts, cross-store promotions, red envelopes, subsidies, livestreaming, cross-platform price comparison. The battle lines grew longer and longer, and the tactics became more and more complex, as if pushing the volume high enough would bring in more consumers and then find growth from a larger transaction scale.

But the market will not always cooperate with this imagination. According to the data listed in the original text, from January to August 2026, national online retail sales of goods grew 4.3% year on year, with the growth rate of related physical goods categories even lower; the growth rate of total online transaction value during this year's 618 also slowed markedly. Behind the numbers is a reality the e-commerce industry is increasingly familiar with: the traffic dividend has not disappeared completely, but it is no longer something you can scoop up just by reaching out as in the past.

Consumers are better at doing the math, and merchants care more about input-output. Subsidies can make people place orders in the short term, but they do not necessarily buy long-term loyalty; a massive marketing campaign may produce lively sales reports, but it may not leave healthy profits. As the ceiling on growth gradually becomes visible, continuing to charge outward is no longer the only answer.

What Tmall sees is that brand users need lower decision costs when facing complex products; what JD.com sees is that fulfillment efficiency and service certainty remain its strengths; what Douyin sees is that there is still a break between content heat and actual transactions; what Pinduoduo sees is that industrial belts still need more accurate matching of supply and demand; what Xiaohongshu sees is that once trust is eroded by excessive marketing, it is very hard to buy back with a single campaign.

This also means that Double 11 is turning from a contest of "who can make more noise" into a business stress test lasting several weeks. Platforms need to look not only at how many people place orders at a given moment, but also at whether products are chosen correctly, whether inventory is misjudged, whether fulfillment can keep up, whether merchants can make money, and whether users will come back.

In the past, when people said "accumulate grain," it was easy to think of cash, inventory, and resources; today, what is truly worth stockpiling also includes high-quality transaction data, stable merchant relationships, effective operational links, and service experiences consumers are willing to use repeatedly. Excitement can be manufactured, but operational capability can only be accumulated bit by bit.

03 Delay claiming kingship: whoever grabs the entrance first is not necessarily the one who laughs last

Why still "delay claiming kingship"? The story from the late Yuan dynasty may offer an interesting reference.

Chen Youliang had strong troops and horses and put his ambitions on the table very early; Zhang Shicheng occupied the wealthy Jiangsu and Zhejiang regions and lacked neither money nor grain, yet gradually became trapped in his own comfort zone; Zhu Yuanzhang did not start with an advantage, but he was in no hurry to reveal all his cards at once. He first established a firm foothold, then judged who was the most dangerous opponent before him, and only when the time was ripe did he gradually complete unification.

Of course, business competition is not a historical war, and platforms are not ancient warlords. But this story at least reminds us: possessing the strongest weapon for a time does not mean one has already mastered the conditions for final victory. Technological leadership can create opportunities, but it cannot replace organizational capability, supply efficiency, user trust, and business discipline.

Over the past two years, there has been a strong sense of urgency in AI competition: if I do not do it first, someone else will; if I do not grab the entrance, someone else will take it. Such anxiety is not entirely unreasonable, but once everyone treats "grabbing the entrance" as the only goal, competition easily turns into duplicated construction and high-cost consumption. Everyone is chasing the same story, but may forget what they originally made money from.

Last year's food delivery war already showed the industry the cost of this approach. Subsidies, traffic, and the fight for entrances can create momentum in the short term, but they do not necessarily buy sustainable advantages. When competitors can all follow with subsidies and consumers can turn to the next one at any time, it is worth recalculating whether the money spent bought loyalty or just a few days of excitement.

This Double 11, platforms seem to be pulling their attention back: rather than rushing to announce that they will rewrite the industry, it is better to first do well the hardest, most expensive, and most error-prone parts of the existing business. Rather than simply manufacturing impulse, it is better to manage certainty. The former is fighting for other people's users; the latter is giving one's own users a reason to stay.

This does not mean platforms will stop competing with one another. On the contrary, competition may become more granular, deeper, and harder to sum up with a single slogan. What will be contested in the future is not necessarily whose AI assistant is best at chatting, but who can understand user needs more accurately, deliver goods more promptly, reduce supply chain waste further, and protect authentic content better. On the surface, everyone is doing AI; looking deeper, everyone is competing for the parts of their own business that others find hard to replicate.

Of course, reconciliation also has a cost. When AI turns from a "disruptor" into a "builder," it is no longer so easy to tell a grand story exciting enough for everyone. The capital market will ultimately still ask: How much has efficiency improved? How much have costs fallen? Has conversion improved? Are users staying? If the answers cannot withstand the test of operating data, even the most beautiful AI narrative will fade.

English text is being prepared.

More than six hundred years ago, Zhu Yuanzhang relied on "build high walls, accumulate grain widely, and delay claiming kingship" to wait for his opportunity in an era when many contenders rose together. Today's e-commerce platforms certainly do not need to copy the ancient art of war, but at a time when traffic growth is slowing, price competition is fierce, and consumers are increasingly rational, these nine characters still have practical meaning.

After all, technology is never the answer itself. Whether technology can be turned into one's own capability is the answer.