Xiaomi's Next Stop: Can Cars and AI Support Its Valuation After Phones?

On March 24th, Xiaomi Group released its annual performance report for 2025. The company's total annual revenue reached 457.3 billion yuan, a year - on - year increase of 25%. The adjusted net profit was 39.2 billion yuan, a year - on - year increase of 43.8%.

Behind these figures, the company is undergoing a profound transformation in identity: from a consumer electronics company centered around mobile phones to a hardcore technology enterprise spanning the three major fields of mobile phones, automobiles, and AI.

01 The growth of the mobile phone business has slowed down, and automobiles have become the second growth curve.

Mobile phones are Xiaomi's core business. In 2025, the revenue of the company's mobile phone × AIoT business reached 351.2 billion yuan, a year - on - year increase of 5.4%, accounting for 76.8%. The mobile phone × AIoT business remains the main source of revenue, but the growth rate has slowed down to single - digits.

Among them, in 2025, the revenue of Xiaomi's smartphones was 186.4 billion yuan, a year - on - year decrease of 2.8%. The shipment volume was 165.2 million units, a year - on - year decrease of 2.0%. The ASP (Average Selling Price) was 1,128.7 yuan, a year - on - year decrease of 0.8%.

The simultaneous decline in revenue, shipment volume, and ASP of Xiaomi's mobile phone business in 2025 was mainly affected by the decrease in shipments in the Indian market. Although there was growth in the Chinese mainland, Latin American, and African markets, it could not fully offset the decline.

In 2025, the sales volume of Xiaomi's high - end mobile phones in the Chinese mainland accounted for 27.1%. The market share in the price range of 4,000 - 6,000 yuan was 17.3%, and the market share in the price range of 6,000 - 10,000 yuan was 4.5%, both showing year - on - year increases.

However, it is worth noting that Xiaomi's overall ASP (Average Selling Price) of smartphones still dropped from 1,138 yuan to 1,129 yuan. This indicates that the shipments of low - end models in emerging markets have dragged down the average price.

Looking at the automotive aspect, in 2025, the revenue of Xiaomi's automotive business was 103.3 billion yuan, and the delivery volume was 411,000 units, a year - on - year increase of 200.4%. For the first time, the operating income turned positive (900 million yuan).

It should be noted that the "positive operating income" of this segment does not include the allocation of headquarters - level expenses such as R & D and sales. The company did not separately disclose the R & D investment in the automotive business for the whole year, but the company's overall R & D expenditure increased by 37.8% to 33.1 billion yuan. One of the main reasons was the "increase in R & D expenses related to innovative businesses such as smart electric vehicles and AI". This segment is still in a stage of simultaneous scale expansion and continuous investment.

In March 2026, the new - generation SU7 series was launched. The number of locked orders exceeded 15,000 in 34 minutes and 30,000 in 3 days. In the sedan market above 200,000 yuan, the SU7 series ranked first in sales in 2025.

However, behind the figures are real challenges. In June 2025, the second model, YU7, was launched. The number of locked orders reached 240,000 within 18 hours of its launch, setting a new industry record.

However, the official website shows that the delivery cycle of the popular configuration of YU7 is as long as 35 - 38 weeks. This means that consumers who place orders now have to wait nearly a year to get their cars. Lei Jun promised to "accelerate production". The annual production capacity of 150,000 units of the first phase of Xiaomi's intelligent manufacturing industrial base has been occupied by SU7, and the mass - production time of the second - phase project has not been determined.

Beyond production capacity, quality control issues have also begun to surface. In May 2025, the SU7 Ultra caused rights - protection actions by car owners due to "functional shrinkage of the carbon - fiber front hatch" and "horsepower being locked". The collision accident in the intelligent driving mode in April also raised doubts about its technological maturity.

To cope with the delivery pressure, Xiaomi launched a "ready - to - buy car" channel in December 2025, including cars converted from abnormal orders, showroom cars, and quasi - new cars with minor defects into sales, and pricing them differently according to the vehicle condition. There are still different views in the market on whether this measure, which is interpreted by the outside world as a "rescue" move, is a good way to ease the supply - demand contradiction or just a stopgap measure to cover up the short - board of the supply chain.

02 Forward - looking layout of AI and robots, difficult to contribute revenue in the short term

In May 2025, Xiaomi released its self - developed 3nm flagship SoC chip, "Xuanjie O1". Behind this is a long - term accumulation.

From the setback of Surge S1 in 2017 to the restart of the "Xuanjie" large - chip project in 2021, Xiaomi has invested more than 13.5 billion yuan in the past four years and assembled a R & D team of more than 2,500 people.

Lei Jun admitted in his annual speech: "If Xiaomi wants to become a great hardcore technology company, chips are the peak we must climb and a tough battle we can't avoid."

Currently, Xuanjie O1 has been installed in the flagship mobile phone Xiaomi 15S Pro and Xiaomi Pad 7 Ultra. At the same time, the first watch chip Xuanjie T1 with self - developed 4G baseband was also released, marking an important step for Xiaomi in the key baseband track.

From the application level, self - developed chips bring not only supply - chain security but also the initiative in product iteration rhythm. While most domestic mobile phone manufacturers are still waiting for the chip release cycles of Qualcomm and MediaTek, Xiaomi can arrange the rhythm according to its own product planning.

However, the long - term iteration and ecological construction of chips is a marathon. Having self - developed chips is one thing, and whether it can be continuously iterated and whether a matching software ecosystem can be built is a more long - term test.

Early in the morning on March 19, 2026, Xiaomi released three large models in one go: MiMo - V2 - Pro, MiMo - V2 - Omni, and MiMo - V2 - TTS. This release rhythm itself reveals Xiaomi's emphasis on the AI strategy.

MiMo - V2 - Pro is a flagship base model for the Agent era, with a total of more than 1 trillion parameters (42B active parameters) and supporting a context length of 1M. It ranks eighth globally and second in China on the Artificial Analysis global large - model comprehensive intelligence ranking list, and scored 86.7 points in the code - ability evaluation SWE - bench Verified.

What's more worth noting is its pricing strategy: the API pricing is only one - fifth of that of the peer competitor Claude Opus 4.6. This means that Xiaomi hopes to attract developers and enterprise users with a lower usage threshold.

MiMo - V2 - Omni is a full - modality base model that integrates text, vision, and voice modalities in a unified architecture at the bottom, breaking the limitation of traditional models that "emphasize understanding and neglect execution".

MiMo - V2 - TTS is a large voice - synthesis model that has been pre - trained with hundreds of millions of hours of voice data and supports multi - granularity emotion control and various dialects.

Xiaomi's investment in AI is not short - term. The company announced that it will invest at least 60 billion yuan in the AI field in the next three years, and the investment related to AI and embodied intelligence in 2026 alone exceeded 16 billion yuan. In 2025, the R & D expenditure was 33.1 billion yuan, and a considerable proportion was invested in AI.

Currently, these models have been launched on applications such as Xiaomi miclaw (mobile - phone - end AI Agent), WPS Lingxi, and Xiaomi Browser, and the API interface has been opened. However, the commercialization path of the AI business still needs to be verified. The high R & D investment suppresses short - term profits, and whether the revenue scale of API calls can cover the costs and form a sustainable business model is the key point to be observed in the future.

Xiaomi has been laying out in the robotics field for five or six years, and a key breakthrough was achieved in 2026. In February, Xiaomi open - sourced the robot VLA large model Xiaomi - Robotics - 0 (4.7 billion parameters) and released the tactile - driven fine - grasping fine - tuning model TacRefineNet for robots. Lu Weibing said: "Robots have just entered the factory, and it will still take time for large - scale application. We think this time may be in the next five years, and it's a bit early now."

From the perspective of industry trends, automobile manufacturing has become the forefront battlefield for the commercialization of humanoid robots. Automobile companies such as BMW, Toyota, Mercedes - Benz, and Audi all have similar cooperation or pilot plans. Morgan Stanley predicts that by 2050, the market scale of humanoid robots may reach 5 trillion US dollars. Xiaomi's choice to enter from automobile factories is not only the best test field to verify technology but also to accumulate data and experience for future large - scale application.

03 Conclusion

Connecting the threads of mobile phones, chips, automobiles, AI, and robots, we can see Xiaomi's strategic logic: using high - intensity R & D investment as fuel, self - developed chips and AI large models as the underlying technology engines, deeply empowering the "full ecosystem of people, cars, and homes", and achieving a leap from technology catch - up to experience leadership in each product line.

Lei Jun announced that he will invest another 200 billion yuan in R & D in the next five years, showing his determination. But as he said, "True gold stands the test of fire." Fighting on the three major tracks of mobile phones, automobiles, and AI simultaneously means a high degree of resource dispersion and high - difficulty management coordination.

Whether Xiaomi can truly become the "global leader of the new generation of hardcore technology" still needs to be tested by time and the market. This company, which started with the Internet model, is answering a fundamental question with down - to - earth R & D investment: how far can it go in the field of hardcore technology.

This article is from the WeChat official account "Big Bowl Thinking". The author is Big Bowl. It is published by 36Kr with authorization.