Meituan's "Defense Ledger": The Aftershock Following the 80 Billion Yuan Subsidy War

After a year-long food delivery subsidy war that cost hundreds of billions of yuan, Meituan finally stabilized its position in the second half of 2026.

Judging from the recently disclosed financial data and institutional estimates, Meituan's operating profit is undergoing a rapid recovery. In the core local commercial sector, the operating profit rebounded from a loss in the third quarter of 2025 (with an operating loss of 140.71 billion yuan in that quarter) to the level of tens of billions of yuan in the second quarter of 2026. The per-order UE (Unit Economics) of food delivery has also broken away from the negative value mire at the height of the subsidy war and returned to a slight profit state. At the recent industry conference, Xue Bing, General Manager of Meituan Delivery, also set the tone accordingly: the proportion of low-price orders has halved, and the order value of merchants is rebounding.

The figures are impressive, but when placed in the current competition coordinate system of China's Internet industry, Meituan is far from reaching the point where it can rest easy. Although Meituan has defended its basic market with extremely strong resilience in this war of attrition, it has also exhausted the "borderless expansion" narrative that it relied on to maintain high valuations in the past decade.

Now Wang Xing is no longer faced with the problem of how to tell a more grand new story, but how to use extremely limited resources to plug the leaking gaps in its food delivery, in-store and AI businesses.

"Improving quality and reducing quantity" of food delivery: Involuted competition after order volume peaks

The biggest sequela brought by the subsidy war to Meituan is that the food delivery market has changed from "one superpower with multiple strong players" to "two strong players facing off".

Taobao Flash Shopping has poured real money into the market to create tens of millions of daily orders within a year, and JD Delivery has also taken a slice of the high unit price market. Although Meituan has stabilized nearly half of the market share and an extremely high repurchase rate by giving up low-price orders and focusing on high-value users, the price paid is a slight drop in the average daily order volume of food delivery.

To fill the gap caused by the loss of low-price orders, Meituan's solution is "improving quality and reducing quantity" — using higher unit price to maintain total GMV and profit. Recently, Meituan has launched a series of tools called "Efficiency Triangle" for merchants, trying to help merchants generate profits on their own by optimizing store formats and AI systems (such as the Miaoshou Food Delivery version).

This is not a sophisticated strategy, but a typical stock game. When the growth rate of orders on food delivery platforms hits the physical ceiling, Meituan can no longer increase profit margins by squeezing the commission from merchants as it did in the past (merchants are already overburdened), so it can only settle for the second best and try to increase service revenue by providing digital tools.

But the ceiling of this business is extremely low. Compared with direct commission collection, the scale of tool subscription and SaaS service fees is negligible. A more realistic question is: at a time when Douyin is continuously penetrating the local life sector with its huge public domain traffic, why should merchants believe that Meituan can bring better increments than Douyin?

Blood loss of in-store business: The profit cow is dethroned

Food delivery is the core traffic pool of Meituan, but it is the in-store travel and hotel business that really contributes the core profit to Meituan.

In the past, the operating profit margin of this business remained above 40% for a long time, which was the confidence for Meituan to use the profit from food delivery to subsidize new businesses. However, the strong rise of Douyin is systematically undermining this profit model.

According to institutional statistics, the GMV of Douyin's local life payment exceeded 850 billion yuan in 2025, a year-on-year increase of about 59%, which is close to the scale that Meituan has cultivated for more than ten years. Under the dimensionality reduction attack of Douyin's traffic, Meituan was forced to follow short-video gameplay such as live streaming sales and influencer store visits, and greatly increased subsidy efforts. The operating profit margin of the in-store business has dropped from 45% at its peak to 26.5% recently.

The meager profit of per-order UE for food delivery, coupled with the continuous pressure on the profit margin of the in-store business, means that Meituan no longer has sufficient cash flow to feed new businesses. This is also why Meituan is extremely cautious in its overseas expansion (Keeta) and has decisively shrunk its community group buying business (Meituan Youxuan). Against the background of eroded core profit pools, the fault tolerance rate for any new business trial and error has been minimized.

LongCat and AI: The "shovel" for defense

In the AI large model battle that determines the future, Meituan has also taken frequent actions. Recently, Meituan has intensively open-sourced the LongCat series of models, some of which have parameter scales of up to trillions, and are supported by a huge domestic computing power cluster.

On the surface, it seems that Meituan is going all out on the general large model track. But stripping away the coat of technical parameters, Meituan's AI strategy is extremely pragmatic, and even can be described as lacking imagination.

Wang Xing made it clear internally that Meituan does not intend to be a "Token factory". This means that the core purpose of LongCat is not to seize the entrance of C-end chatbots, nor to make huge profits by selling large model APIs externally.

The underlying logic of Meituan's AI development is "AI at Work". In food delivery fulfillment, AI is used to optimize scheduling algorithms and predict food preparation time; on the merchant side, AI is embedded in the Miaoshou system to help merchants design menus and handle customer complaints.

This logic is completely correct, which is in line with Meituan's gene of heavy assets and heavy offline operation. But it also illustrates a cruel fact: Meituan's AI is not a "flywheel" for outward expansion, but a "shovel" for inward efficiency mining. Unlike ByteDance's Doubao or Alibaba's Tongyi, it cannot create a brand new business ecosystem, and can only be used to defend the profit margin of existing businesses.

In addition, the personnel changes in Meituan's core AI team (such as the rumor of the departure of core personnel of the LongCat large model, which Meituan has denied) also add uncertainty to this defensive battle. At a time when computing power costs are high and general large models are iterating extremely fast, whether completely giving up the catch-up of general capabilities and only making fine-tuning in vertical scenarios will be subject to dimensionality reduction attack in the long run is an unresolved hidden danger.

Conclusion: From "borderless expansion" to "bordered defense"

Meituan's experience is a classic epitome of China's large Internet companies after bidding farewell to the era of high growth.

In the era when mobile Internet dividends exploded, Wang Xing believed in "borderless expansion", and Meituan could fight on multiple fronts at the same time, including food delivery, hotel and travel, ride-hailing, shared bicycles, and community group buying. As long as capital is still circulating and traffic can still grow, scale can overwhelm everything.

But now, the order volume of food delivery has peaked, the in-store business has been caught up by short-video giants, and the cash cow of new businesses has lost significant profits. The only weapons Meituan has to fend off competitors are the profits squeezed out through cost control, iterated tools, and the extreme control over the instant delivery network.

Meituan is still a very combat-effective company. Being able to stabilize its basic market in the meat grinder of tens of billions of subsidies is enough to prove its organizational capability. But it has to accept a reality: the capital market no longer regards it as an "expansion platform" with unlimited growth elasticity, but prices it as an "efficiency-oriented company" with under pressure profits and struggling to defend in the red ocean.

In this new cycle, the quality of defense is far more important than how good the story is told.

This article is from the WeChat official account "Tech Business", authored by Tech Business, and published with authorization from 36Kr.