Xiaomi made 39.2 billion, Lu Weibing said, "Let's hold on first."
```
Author | Zhou Zhiyu
In March 2021, Xiaomi CEO Lei Jun announced the company's entry into the car business, calling it “the last major entrepreneurial project of his life” and expressing his willingness to stake his entire reputation on it. At that time, Xiaomi had just rebounded from a low point in its smartphone business, and the market was skeptical about a phone maker branching out into car manufacturing.
Four years later, on the evening of March 24, Xiaomi Group released its full-year 2025 results. Annual revenue from the automotive business surpassed 100 billion, and the segment's operating profit turned positive for the first time.
Counting from the first batch of SU7 deliveries in April 2024, Xiaomi took less than two years to complete the journey “from zero to 100 billion in revenue, from loss to profit”—Li Auto took more than four years, and NIO only reached quarterly profitability in Q4 2025 after eleven years since its founding.
But this annual report carries two sentiments. On the whole, it was the best year in Xiaomi's history, but the fourth quarter was under pressure, with gross margins for both smartphones and cars turning downward. That evening at the earnings conference, Xiaomi President Lu Weibing said: “The start isn't easy, and there will be some short-term performance pressure.”
Xiaomi's situation is not unique. 2025 is a rare bumper year jointly experienced by China's technology and new energy industries—vehicle purchase tax was fully waived, state subsidies were robust, AI large models landed collectively, and penetration of new energy continued to climb rapidly. Almost all major players delivered their best data ever.
Now, almost all players need to answer this question: When these favorable conditions cease to exist simultaneously in 2026, who can withstand the headwinds?
The Two Sides of Profitability
The core highlight of Xiaomi's financial report is the change in its revenue structure.
The automotive division’s gross profit for the year was 25.8 billion yuan. The phone division’s gross profit was 20.3 billion yuan. A business less than two years old already exceeded the gross profit made by the phone business of fifteen years. Annual revenue reached 457.3 billion yuan, with adjusted net profit of 39.2 billion yuan, both historical highs. Automobile deliveries were 411,082 units, a year-on-year increase of 200%.
That's one side.
The other side is Q4. The adjusted net profits of the previous three quarters were 10.7 billion, 10.8 billion, and 11.3 billion, showing a consistent upward trend. In Q4, it dropped to 6.3 billion, down 23.7% year-on-year. The Q4 smartphone gross margin was only 8.3%, car gross margin dropped from 25.5% in Q3 to 22.7%, and IoT went from 23.9% to 20.1%. All four lines declined in Q4.
The 8.3% for phones requires elaboration. Lu Weibing previously disclosed to Wallstreetcn that memory prices rose, with 12+256GB memory increasing from around $30 at its low point to about $130—nearly quadrupling. Now he’s revised his estimate upwards, speaking frankly that the increase was “even more aggressive than my already aggressive forecast,” and the cycle “may last longer than we currently judge.”
The higher the proportion of storage costs in the BOM, the greater the impact. Xiaomi’s full-year smartphone ASP was only 1,128 yuan, with Redmi and POCO, the volume drivers, bearing the brunt. High-end models are offsetting the pressure, with the market share for the 6,000 to 10,000 yuan range nearly doubling to 4.5%, but Q4 data shows this wasn’t enough to outpace cost increases.
Competitors have begun raising prices. Lu Weibing said: “I can really understand them, because it's difficult for everyone. We’ll try to hold out for now, but if we really can’t, we’ll also have to raise prices.”
His reluctance is practical: phones don't just make money on hardware. Internet services brought in 28.6 billion in annual gross profit, with a margin of 76.5%, a figure based on shipment scale. If you cut volume to protect margins, you shrink your traffic base.
Automotive gross margin was 24.3% for the year, surpassing BYD, Li Auto, and Great Wall. But Q4 fell to 22.7%, mainly due to a lower proportion of SU7 Ultra deliveries and year-end inventory clearance. Rising storage prices also impacted cars, with Lu saying there's a fair amount of memory in them, though its proportion of the car’s BOM is lower than in phones.
A more crucial calculation: Automotive gross profit was 25.8 billion, operating expenses 24.8 billion, leaving an operating profit of just 900 million. 106.1 billion in revenue yielded an operating margin of less than 1%. Xiaomi CFO Lin Shiwei emphasized that this segment is fully named "Intelligent Electric Vehicle and Innovative Businesses such as AI"—AI investment is increasing, and new businesses are still in the investment period.
Betting on the Next Cycle
Where did the profit go?
Last year, Xiaomi repurchased a total of 6.3 billion HKD. R&D was 33.1 billion, up 37.8%. Capital expenditure was 18.2 billion, with 12 billion for the automotive division, triple the year before. Inventory rose from 62.5 billion to 80.9 billion.
In 2026, investment will continue. Lu Weibing confirmed that this year’s R&D will exceed 40 billion, with 16 billion for AI, and over 60 billion for AI over the next three years. The annual report gave an even bigger figure: from 2026 on, cumulative R&D will exceed 200 billion over five years.
BYD’s full-year R&D exceeded 50 billion, Huawei has been above 100 billion for years. Xiaomi is using profits to buy an admission ticket to the next cycle.
Lu Weibing revealed that this year, a product featuring both the self-developed chip Vela O1, HyperOS, and a self-developed AI large model will launch—he calls this the "grand convergence." 13.5 billion has already been invested in Vela O1, with a team of 2,500 people. “A chip is essentially a platform capability; with the platform, you can make many products.”
He repeated this view at the earnings conference. The MiMo-V2-Pro large model exceeds one trillion parameters, ranking fifth globally by brand. The smartphone-end Agent miclaw is under closed testing, and Lu defines it as “the prototype of an Agent-era operating system.”
But commercialization of AI? He admitted that it still seems a bit early.
Another direction for profit reinvestment is expansion overseas. The report shows revenue share from mainland China rose from 58.1% to 67%, while overseas declined slightly. Cars pushed up the domestic share, but also increased dependence on a single market. Xiaomi needs overseas growth.
Lu told Wallstreetcn an even more complete overseas roadmap than in the annual report. Excluding North America, the IoT market abroad is three times the size of the domestic market, but Xiaomi’s overseas IoT business is only one-sixth that of China's. “If we reach domestic levels, that's potentially sixfold growth.”
Lu believes 2025 is the first year for major appliances to go overseas, covering four major regions and fourteen countries, with integrated delivery and installation services now rolled out in several places. He visited the Xiaomi Home in London: “They mainly sell high-end products, even the rice cookers sell very well.” There were 450 overseas new retail stores at the end of last year, aiming for over 1,000 by year-end.
High-end phones are also going global. The Xiaomi 17 series, released in Barcelona in February, starts at 999 euros overseas; the Leitz Phone (the Xiaomi 17 Ultra Leica version) is priced at 1,999 euros—already above the iPhone 16 Pro Max in Europe. Lu said: “Today Xiaomi has the courage and confidence to challenge the iPhone at higher price points.”
The car business is last to go overseas, but most anticipated. Entry into Europe is set for 2027: hard before easy, and Lu is directly in charge of the overseas preparation group.
After research trips to Europe, Xiaomi insiders concluded that now is an excellent time for Chinese cars to go overseas. Even before official European entry, there are already European users importing Xiaomi cars themselves, getting them licensed and registered, with brand recognition taking shape.
How does the market price all this? Despite Xiaomi's complex business, it’s not hard to calculate. From the Q3 peak until now, Xiaomi's share price has nearly halved. For a simple calculation: in 2025, the adjusted net profit of the smartphone × AIoT division will be about 38 billion, with a valuation close to Xiaomi's own historical average of about 20-22x PE. The Automotive & AI division with 106.1 billion in revenue gets 1.4-1.8x PS.
What’s inside the automotive division is not just cars—it includes large models, Agent, robots, whole-house smart, and 1 billion IoT devices with 750 million monthly actives. If you price the auto business at just 1x PS, the value of AI and overseas expansion is pretty much invisible.
Judging from Xiaomi's share price, clearly investors haven’t given much premium.
This reflects two overlapping worries. One, profits in 2026 may be further downgraded, memory prices have not peaked, state subsidies are just starting to fade, and whether Q4's 6.3 billion is the new normal is unclear. The other is market uncertainty: when will the reinvested profits pay off? The “grand convergence” of chip + OS + large model comes out this year, but what increment can it bring? The prospect of 6x overseas IoT growth sounds alluring, but 1,000 stores are just the beginning. Overseas car sales start in 2027. Every card points to the future, but none can be played immediately.
The five-year 200 billion R&D plan is fundamentally a bet—when the memory cycle is over and the industry reshuffle is done, Xiaomi not only remains at the table, but also holds more cards than others.
The Wind Will Not Always Be at Your Back
Put Xiaomi’s Q4 in the context of the industry.
Over the past decade, China’s new energy industry has gone through a long process of validation. Around 2015, a wave of new players emerged, surviving on financing. In 2018-2020, the elimination began: Byton, Saleen, and others went under. From 2022 to 2023, HiPhi halted, Neta was in peril, Nio lost over 20 billion a year, and William Li’s keyword was “survive.”
By 2025, the situation reversed. Nio and XPeng made a quarterly profit for the first time in Q4, Leapmotor was profitable for the full year. Xiaomi went from first delivery to annual profit in less than two years. Almost all surviving top players wrote their best pages this year.
But 2025 was also the peak of all favorable conditions. Car purchase tax was fully waived, the best-ever subsidies, the fastest penetration increase. In 2026, these factors won’t all be present. Subsidies halve, capacity keeps growing—Xiaomi at 550,000, Leapmotor at 1 million, BYD and Huawei price competition across all segments.
Memory price increases are a hidden thread connecting the new energy and consumer electronics tracks. AI’s surging demand for HBM has squeezed out consumer capacity; some low-capacity types are even unavailable or discontinued.
Besides storage, the price rises of all kinds of materials are pressuring every player in the field.
Lu Weibing said, “After the price hike cycle ends, I believe many categories will see a reshaping of industrial structure. Some companies may face extreme operating difficulties in such a long price-upcycle and may even suffer huge losses or fail to survive.”
He added: “What we should start paying attention to is what comes after the price-upcycle.” His judgment: “A harsh external environment will inevitably force a lot of innovation.”
Reshuffling means that the lower a manufacturer is on the high-end ladder, the higher memory cost is as a proportion of BOM, and the greater the pain. After this storage cycle, mid-tier players in consumer electronics will be cleared out further.
At the same time, the story of tech companies building cars has entered its second act. The first act took ten years, solving “Can it be built, sold, and turned profitable?” Xiaomi and Huawei both answered yes with their 2025 data. The second act asks different questions: with subsidies receding, costs spiking, and competition intensifying, can this model survive a full downcycle?
Lu Weibing said bluntly at the earnings meeting: “This year’s challenges are very, very big.”
He added: “But I believe through our team’s efforts in 2026, we can still deliver a report card we find acceptable.”
The words “acceptable” are those used only by an executive who has been through cycles. This may be the truest undertone of 2026.
Risk DisclaimerThe market has risks, and investment must be prudent. This article does not constitute personal investment advice nor does it take into account any individual user's particular investment objectives, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article are applicable to their specific circumstances. Invest at your own risk. ```