Xiaomi Steps Up on "Safety Lessons"

```

Author | Zhou Zhiyu

Editor | Zhang Xiaoling

On the first day of the Guangzhou Auto Show, Lei Jun did not appear at Xiaomi Automobile's launch event.

Standing on stage was Xiaomi Automobile Vice President Li Xiaoshuang. Aside from technical iteration, the core of Li Xiaoshuang's speech had only one focus: safety. On the screen were three large lines without any small annotations—Safety is the premise, safety is the foundation, safety is everything.

The marketing tone suddenly faded, and the safety card was played. This year, as Xiaomi has been deep in the vortex of public opinion, some subtle changes are taking place.

At the Guangzhou Auto Show, Li Xiaoshuang did not talk about fancy add-on features, but focused on showcasing the Xiaomi HAD Enhanced Version and safety-assist features.

In the first half of intelligent driving, the industry generally adopts "rule-driven" or basic "data-driven" approaches. It’s like making drivers memorize traffic laws or learning to drive through imitation and memory. But this model quickly hits a ceiling: when facing extremely rare scenarios, due to the lack of training samples, models cannot effectively learn.

The "world model" that Xiaomi is promoting this time essentially builds a high-fidelity virtual simulation engine. In this virtual world, the system no longer relies on accidents in real life to learn, but instead can conduct countless simulations and error-trials in generated scenarios— “score points for correct moves, deduct points for mistakes.”

This paradigm shift from “imitation” to “cognition” aims to use algorithm generalization to cover long-tail risk scenarios, thereby reducing systemic driving risk.

This time the Xiaomi HAD Enhanced Version AEB has been comprehensively upgraded: its forward speed domain has been expanded from 5-135km/h to 1-135km/h, and a new rear speed domain of 1-30km/h has been added; Xiaomi has also extended low-speed collision avoidance (L-AEB/R-AEB) identification objects from just traditional vehicles and pedestrians, to water barriers, anti-collision barrels, pillars, and even walls. In response to the industry-wide problem of "misstepping on the accelerator," Xiaomi has introduced MAI (Mis-acceleration Inhibition) linked with the underlying AEB.

Behind the densely released functions, Xiaomi is trying to use technology to build a firewall. Li Xiaoshuang would also “patch up” as needed. He emphasized that assisted driving is not autonomous driving, and the driver still needs to stay focused at all times.

Right before the Guangzhou Auto Show (November 18), Xiaomi delivered its "strongest quarterly report ever"—for the first time, Xiaomi’s automotive business achieved single-quarter operational profit, with quarterly deliveries exceeding 100,000 units. However, the capital market gave a surprising response: the day after the results announcement, the stock price did not rise but fell, even showing dramatic fluctuations.

Behind the huge volatility in stock price lies the deepest anxiety the capital market has about Xiaomi Automobile: when the traffic dividend reaches the ceiling, when 500,000 cars are on the roads, can this “internet celebrity car company” bear the backlash from safety incidents?

Since the release of SU7 on March 28, 2024, it took Xiaomi Automobile just over 600 days to roll out its 500,000th complete vehicle. This speed is rare in world automotive history. In the early stages, the capital market cares about speed and the ability to produce blockbusters, and is willing to assign a high premium for it.

But at the 500,000-unit level, the logic changes.

With a scale of 10,000 units, occasional accidents can be digested by fans' filters and other means; but with a base of 500,000 units, even a low-probability safety flaw, multiplied by the massive number of cars, could become a "black swan" event that shakes the brand’s foundations.

For Xiaomi, with its huge “traffic halo,” it can endure a “safety trust crisis” less than any traditional car company.

After the Tongling and Chengdu accidents, public scrutiny of founder remarks showed that entrusting the brand’s safety to a single individual is not a long-term solution.

At this Guangzhou Auto Show, Li Xiaoshuang’s appearance and the unveiling of the 1,800-person engineering team were also Xiaomi Automobile’s attempt to showcase its maturity in organizational structure. It seeks to communicate to the outside world: Xiaomi is no longer just a startup driven by the founder’s charisma, but a mature car company with a well-established bureaucracy, professional technical teams, and industrial heritage.

This transformation is crucial for the battles to come. Though Xiaomi YU7 already took the top spot in SUV sales for October, it's a model aimed at “family users.” Unlike coupe users, family users are highly sensitive to safety and will not tolerate any internet-style flamboyance.

Only when Xiaomi can prove that even without Lei Jun's hype, it still can rely on its organized engineering team to build the safest and most reliable cars, will it actually win over this most conservative and largest market.

At the Guangzhou Auto Show, Xiaomi has become much less sexy.

It no longer discusses those details that make female users scream, nor does it create viral hits on short video platforms. But this lack of excitement is exactly what the capital market wants to see now.

After crossing the threshold of 500,000 units, Xiaomi Automobile no longer needs to prove how popular it is, but needs to prove how stable it is. The string of accidents and public opinion crises throughout this year have been wake-up calls, alerting Xiaomi still basking in the "miracle of delivery."

When the marketing bubble fades, Xiaomi must prove that even without Lei Jun’s hype, it can still use its industrial capacity to build reliable cars. This is a catch-up lesson that must be completed for survival.

Risk Warning and DisclaimerThe market involves risk, and investment should be cautious. This article does not constitute personal investment advice and has not taken into account individual users' particular investment objectives, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions herein are suitable to their specific circumstances. Investment based on this is at your own risk. ```