The autumn 2026 earnings season amounts to little more than a midterm exam.
On an evening before the 2025 Spring Festival, at a restaurant somewhere in Beijing, Liu Qiangdong treated Wang Xing to a meal, during which Liu Qiangdong said, "Brother, I will formally enter the food delivery business."
This sentence is like a stone thrown into a lake. Four months later, Liu Qiangdong wore a red rider uniform and delivered several orders of takeout on the streets of Beijing, and at night, he had hot pot with about 40 to 50 riders. He raised his glass and said, "I also made a point to deliver a few orders of takeout this afternoon."
At the time, no one thought the war would last so long, and no one expected that by the fall of 2026, a year and a half later, when the three platforms simultaneously released their latest financial reports, the definition of "victory" would have undergone a fundamental change.
On August 28, Meituan released its Q2 2026 earnings report, with core local businesses achieving operating profits of 5.67 billion yuan and a profit margin of 7.9%. In the previous quarter, this figure was a loss of 2.03 billion yuan.
For Wang Xing, this "stop-loss report" has arrived not a moment too soon. In the third quarter of 2025, Meituan's core local business had recorded a loss of 14.07 billion yuan - the most brutal moment of the war. Now, with profits turning positive again, although the 7.9% profit margin is still short of the pre-war level of 40%, the defender's wall has not collapsed.
Eight days ago on August 20, Alibaba released another set of results: China's instant retail revenue reached 53.295 billion yuan, up 45% year-over-year. In the earnings report, Wu Yunming did not emphasize the order peak, but instead focused more on the "continuous improvement in the unit economics of Taobao Flash Purchase", attributing it to "increases in average order value and fulfillment efficiency".
Alibaba's victory does not lie in the total amount, but in the economic model of each order. When JD.com used tens of billions of yuan in subsidies to impact the market, Alibaba took a different approach: instead of competing to see who could offer more, it competed to see who could offer more value.
The first to submit its report was JD.com. On August 13, JD.com's Q2 2026 earnings report showed that its new businesses, including food delivery, incurred an operating loss of 9.85 billion yuan, narrowing by 4.92 billion yuan from the same period in 2025, and by nearly 5.9 billion yuan from the quarterly high of 15.74 billion yuan in Q3 2025. CEO Xu Ran wrote in the earnings announcement, "The path to profitability has reached a clear turning point."
From 15.74 billion to 9.85 billion, JD.com reduced its losses by nearly 40% in just three quarters. However, Xu Ran's tone was cautious - she spoke of a "turning point" rather than "profitability". JD Takeout is still losing money, it's just that the losses have slowed somewhat.
Shifting Gears in Storytelling
What's Left Behind
At the start of the war, scale was the only passport.
In April 2025, Liu Qiangdong laid it out straightforwardly at JD.com's headquarters: "I can always lose money on the front end selling food, as long as I make money on the supply chain." He added, "Now 40% of cross-sales go to our e-commerce platform, and the money we lose is still more cost-effective than buying traffic from Douyin or Tencent."
This is the first narrative: the essence of JD Takeout is not takeout, but an entry point for the supply chain and a customer acquisition channel for e-commerce.
To make that case, JD.com highlighted three figures in its second-quarter 2025 earnings: daily order volume peaked at over 25 million during the 618 shopping festival, more than 1.5 million "quality merchants" had joined the platform, and full-time riders exceeded 150,000. Xu Ran said at the time that food delivery "has successfully achieved our initial strategic goals"—meaning JD had secured a seat at the table.
Liu Qiangdong's account of recovering payments from his supply chain remains to be verified.
Alibaba's shift is more direct.
In August 2025, Wu Yongming said on the earnings call: "We are investing heavily in instant retail, and have quickly achieved initial results, winning over consumers." At that time, Taobao Flash Purchase had been live for four months, and its primary task was to prove it could bring high-frequency consumption back to Taobao.
Three months later, the narrative was fine-tuned to: "The scale of instant retail has expanded, and unit economics have improved significantly." Another three months later: "While maintaining market share, unit economics have continued to improve month-over-month."
Meituan, playing defense, is also seizing the opportunity to reassess itself.
In May 2025, Wang Xing said during an earnings call, "The low-quality, low-price 'involutionary' competition is unsustainable in the long term." This is not a sign of weakness, but a declaration of war - Meituan aims to defend not just orders, but also the quality of merchants, riders, and its fulfillment network.
Six months later, Meituan disclosed that the daily order peak of its instant retail business exceeded 150 million, with an average delivery time of 34 minutes. Wang Xing's tone has become "consolidating core competitiveness and maintaining industry leadership" - after subsidies have been paid out, he wants to prove that orders and network density have not been disrupted by the two attackers.
By the second quarter of 2026, Wang Xing's tone had changed: "Meituan resolutely opposes involution. We will actively cooperate with relevant regulatory investigations. At the same time, in the process of consolidating our market-leading position, we will also reduce resource investment in low-quality orders."
How Much Money Went Up in Flames
Narratives can shift gears, but account books don't lie.
When the observation period is extended, the costs borne by the three companies for this war become clear.
From the first quarter of 2025 to the second quarter of 2026, JD.com's marketing and fulfillment expenses increased by 57.83 billion yuan compared to the corresponding period the previous year; Meituan's logistics and promotional incentive expenses increased by 82.74 billion yuan. From the second quarter of 2025, when Taobao Flash Purchase was launched, to the second quarter of 2026, Alibaba's sales and marketing expenses increased by 95.45 billion yuan.
The three sets of figures all reflect the group's broad definition, including spending on retail, warehousing and distribution, other instant retail operations, overseas expansion, and user acquisition. But even so, nearly 240 billion yuan in incremental costs would be enough to keep any CFO up at night.
The question is: what did this money buy?
Data from the National Bureau of Statistics provides a calm answer. After the fifth national economic census revision, the country's catering revenue was 5,618 billion yuan in 2024, up 5.3% year-on-year; it increased to 5,798.2 billion yuan in 2025, with the growth rate slowing to 3.2%; in the first half of 2026, catering revenue was 2,825.5 billion yuan, up 2.8% year-on-year.
In July 2025, when the subsidy war on food delivery platforms was most intense, national catering revenue reached 450.4 billion yuan, with year-on-year growth of only 1.1%; among them, catering revenue from units above the quota decreased by 0.3% year-on-year.
Platform orders can shift rapidly under heavy subsidies, but overall restaurant demand does not double in tandem. The food-delivery war is essentially a redistribution of market share—the orders JD and Alibaba have taken from Meituan have not led Chinese consumers to eat more meals; they have simply moved the same meal from Platform A to Platform B.
This means that the "improvement" of the three platforms cannot be built on the incremental growth of the catering industry at the same time. If someone wins, someone must lose.
It is likely that more consumers have benefited from the subsidies.
AI Takes Center Stage
Just as the three platforms were revamping their scoring systems for food delivery, a more expensive narrative entered the earnings report center: AI.
Alibaba is the most aggressive, with capital expenditures of 67.68 billion yuan in the second quarter of 2026, up 75% year-over-year, mainly used for AI infrastructure. External commercial revenue of Alibaba Cloud grew 45%, and AI-related product revenue achieved triple-digit growth for the 12th consecutive quarter. For Wu Yunming, AI is not just a cost-cutting tool for food delivery, but a second growth pillar on par with consumption.
JD.com has not disclosed its independent investment in AI, but emphasized supply chain efficiency. In the first half of 2026, JD Industrial deployed over 70 AI intelligent bodies on the procurement and fulfillment chain. Unlike Alibaba's search for a second curve and Meituan's transformation of local life entrances, JD.com is still trying to put AI back into the gears of its supply chain.
All three companies have put AI front and center, but the rules of the ledger have not changed. AI must prove its ability to generate revenue or reduce costs, while food delivery must demonstrate its ability to retain customers, drive cross-purchasing, and generate supply chain profits. Two different narratives about the future ultimately boil down to the same profit statement.
Predictions: Three Possible Strategies for the Second Half
At the one-year milestone of the food delivery war, the next steps for the three platforms may involve three different strategies.
JD.com: The Verification Period for Supply Chain Payments
A quarterly loss of 9.85 billion yuan seems substantial, but Xu Ran's notion of a "turning point" is not entirely unfounded. JD.com's true strength lies not in its front-end food delivery service, but in its backend logistics and supply chain network. If the cross-buying and supply chain efficiency brought about by food delivery can lead to improved profit margins in its retail business over the next four quarters, JD.com's narrative will be validated. Conversely, if retail profits cannot offset the losses from food delivery, the "supply chain returns" narrative of JD.com's food delivery business will face even more scrutiny.
Predictions indicate that JD Takeout will continue to narrow its losses in the second half of 2026, but the annual loss for its new businesses may still exceed 30 billion yuan for the year. The true test will come in 2027.
Alibaba: Tmall's Flash Sale has independent profit and loss accounts
Instant retail revenue of 53.295 billion yuan increased 45% year-over-year, but Alibaba has never separately disclosed the profitability of Taobao Deals. The "sequential improvement" in unit economics is a vague statement - by how much did it improve, and how far is it from breaking even? Alibaba needs to provide a clearer average economics model over the next two quarters, however, the incremental growth brought by food delivery to traditional e-commerce is indeed visible to the naked eye.
Predictions suggest that Alibaba will disclose Taobao Flash Purchase's standalone operational data for the first time in the third or fourth quarter of 2026. If the single-order loss has nearly broken even, Alibaba's instant retail narrative will likely be reevaluated.
Meituan: Profit Recovery and AI Efficiency as Dual Drivers
A quarterly profit of 5.67 billion yuan is a positive sign, but a profit margin of 7.9% still has a significant gap with the pre-war level of 21%. Meituan's path to recovery depends on two variables: one is whether industry competition is truly becoming rational (whether regulation continues to constrain subsidy wars), and the other is whether AI can bring substantial efficiency improvements in rider dispatch, merchant operations, and user recommendations.
Meituan's core local commerce business is expected to see its profit margin rebound to 10%-12% in the second half of 2026, but it will be difficult to recover to pre-pandemic levels within the year. The improvement in fulfillment efficiency driven by AI is expected to start showing up in the financial statements in 2027.
A War with No End
At a dinner before the 2025 Spring Festival, when Liu Qiangdong told Wang Xing "I will formally enter the food delivery market", neither of them probably thought that this war would evolve into a protracted and costly battle lasting a year and a half, with expenditures in the trillions of yuan.
What's even more unexpected is that as the war drags on, the definition of "winning" has become increasingly diverse. JD.com needs to prove that its supply chain can generate returns, Alibaba needs to prove that every order can be accounted for, and Meituan needs to prove that its network can still turn a profit after defending its territory.
All three companies claim they are improving, but improvement is not the same as victory, reducing losses is not the same as turning a profit, and improvements in unit economics do not necessarily mean they are profitable. The food delivery war has no end, it has simply transitioned from the first phase of "burning money to grab market share" to the second phase of "proving the business model".
The earnings season this fall is only the first midterm exam of the second phase, and the real report card may not be revealed until the spring of 2027.
At that time, the story of AI may have already rewritten the entire rules of the game.