The public intelligence has been doused with a bucket of cold water

On March 29th at 10:44 PM, a Xiaomi SU7 standard version was involved in a serious traffic accident in Tongling, Anhui. The accident resulted in the deaths of the driver and two passengers. All three were university students, and they were traveling from Wuhan to Anhui for an exam.

On April 1st at noon, Xiaomi released a statement saying that the vehicle was in NOA intelligent assisted driving mode, traveling at a speed of 116 km/h before the accident. The road section where the incident occurred was under construction, and the vehicle detected an obstacle, issued a warning, and began to decelerate. The driver then took control of the vehicle, continued to decelerate, and steered the vehicle, after which the vehicle collided with a concrete post of the isolation barrier. The last confirmed speed before the collision was approximately 97 km/h.

Traffic accident information from Anhui Highway Travel Service shows that at 11:12 PM on March 29th, the traffic police reported a fire in a sedan heading from Qimen to Zongyang on the DeShang Expressway; by 12:19 AM on March 30th, the open flames had been extinguished.

On the same day, the annual China Electric Vehicle Hundred-Person Forum was being held in Beijing.

Many industry leaders attended the meeting: BYD Chairman Wang Chuanfu, Chery Group Chairman Yin Tongyue, Geely Automobile Group CEO Gan Jiayue, Nio's Li Bin, and Lei Jun, who crossed over from the mobile phone industry to make cars.

Almost all the industry leaders present mentioned the same topic in their speeches: intelligent driving.

As early as February this year, BYD proposed the concept of "Universal Intelligent Driving," shifting the battleground of competition in the automotive industry to the field of intelligent driving, followed by Geely and Chery.

When intelligent driving becomes a standard feature, is it more cost-effective for car companies to develop it in-house or to find third parties to quickly build their own intelligent driving systems? However, under the slogan of universal intelligent driving, what users actually experience is mostly just NOA assisted driving on highways, which is still far from true high-level intelligent driving.

With the Xiaomi car accident igniting public opinion, the safety issues surrounding intelligent driving have also prompted new reflections within the industry: Are we moving too fast with intelligent driving?

All car companies are talking about intelligent driving

If BYD is the absolute king of the mid-to-low-end market, then newcomer Lei Jun has directly inserted a sharp knife into the abdomen of the high-end car market with the SU7ultra.

As the most personable and articulate internet celebrity entrepreneur, Lei Jun's speech at the Hundred-Person Forum brought a sense of harmony to the tumultuous automotive circle. Lei Jun "emotionally" recalled the scene from a year ago:

"Looking back a year ago, many peers expressed their congratulations on the launch of Xiaomi cars, and what particularly moved me at this moment is that at last year's launch event, five founders and chairmen of companies personally attended to show their support, including our Great Wall's Wei, BAIC's Zhang Jianyong, and several of my good friends, Li Bin, He Xiaopeng, and Li Xiang, which really touched me."

Amidst his emotions, Lei Jun boasted about the sales of the recently launched SU7ultra:

"After its release last month, priced at 529,900 yuan, what shocked me was that in less than three days, we sold 10,000 units, completing the annual target ahead of schedule." "Many people may think that the target of 10,000 vehicles is set too low. I checked, and last year, over 500,000 luxury cars sold a total of 121,000 vehicles throughout the year."

At this moment, Li Bin from Nio, sitting in the audience, might not be feeling great. This is because Nio's entrepreneurial vision is to penetrate the territory of BBA (Benz, BMW, Audi) and occupy the high ground of high-end vehicles. However, data from March 10 to March 16, 2025, shows that Nio's sales during that week were 2,000 vehicles, far below those of Li Auto (7,900), XPeng (6,900), and Xiaomi (6,500), among other new car-making forces.

As the competition over sales and prices has escalated to the point where government departments are calling for a "stop," the battleground for the competition kings has shifted to the field of intelligent driving.

At the 2025 Electric Vehicle Hundred People Forum, all the automotive giants present vigorously promoted the intelligence level of their cars and their future heavy investments in intelligence.

BYD Chairman Wang Chuanfu stated, "The second half of the intelligent transformation will take about 2 to 3 years, with the pace of transformation accelerating. Previously, we measured the industry's development speed by the penetration rate of new energy vehicles, but starting this year, we will also use advanced intelligent driving to promote the high-quality development of the industry."

BYD, which has always been considered lagging in intelligence within the automotive circle, suddenly took the lead in the trend of intelligent driving this year.

On February 10 of this year, BYD held a strategic release conference for intelligence at its headquarters in Shenzhen, announcing the "Eye of God" advanced intelligent driving system. Wang Chuanfu stated at the conference, "Today, I officially announce that BYD's 'Eye of God' has opened the era of universal intelligent driving."

Once the concept of "universal intelligent driving" was introduced, other car companies quickly followed suit, with new terms like "intelligent driving equality," "intelligent driving accessibility," "technology equality," and "oil-electric equality" immediately occupying the minds of car enthusiasts.

On March 3, Geely launched the Qianli Haohan intelligent driving system, which has five solutions ranging from low to high: H1, H3, H5, H7, and H9, two more than BYD.

On March 18, Chery launched the Falcon intelligent driving system, which has four systems: 200, 500, 700, and 900, one more than BYD.

Next, the pressure of intelligent driving has shifted to private car companies like Great Wall and foreign-funded enterprises like BBA.

Li Bin: Building an intelligent driving chip requires 1,000 fewer battery swap stations

Intelligent driving is divided into five levels from low to high: L1-L5. L1 and L2 can only be called assisted driving because they require human takeover at any time. L3 is partial automation, where vehicles can drive automatically on closed highways and some urban roads, but human intervention is still needed in emergencies. L4 and L5 are fully automated driving From the user's perspective, the most commonly used intelligent driving features are still L2 level lane keeping, automatic lane changing, adaptive cruise control, etc., and they are generally used on highways, while city roads are not easily attempted.

A user of XPeng stated: "China's urban roads are too complex; a pedestrian may suddenly cross the road ahead, and an electric scooter rider may suddenly appear behind. It's unpredictable, and I dare not hand over the car to the intelligent driving system."

To allow users to confidently hand over their cars to the intelligent driving system on urban roads, automakers still have a long way to go and need to invest significant efforts.

Is the research and development investment in intelligent driving software and hardware done in-house by automakers or outsourced to third parties? Each company has its own calculations.

Nio is a strong advocate of in-house development. At this year's Hundred People Meeting high-level forum, Nio's Li Bin's speech focused on promoting the newly launched ET9. Li Bin stated: "The Shenji NS9031 chip mass-produced on the ET9 is the world's first automotive-grade 5-nanometer high-level intelligent driving chip, developed over more than three years."

For this chip, Li Bin invested heavily, "This chip is equivalent to the investment of 1,000 battery swap stations."

Li Bin revealed: "Nio's R&D investment last year was 13 billion yuan, the year before was 13.4 billion yuan, with an average quarterly investment of 3 billion yuan, totaling nearly 60 billion yuan since its establishment." Among this, the investment in the R&D of the intelligent driving system's software and hardware accounts for a significant proportion.

The advantage of automakers developing their own intelligent driving systems is that they own the user driving data, are aware of system bugs, and achieve seamless integration between the system and the vehicle. The downside is the long R&D cycle and huge costs. If they cannot quickly scale sales to amortize the R&D costs of the intelligent driving system, this heavy burden may very well drag the automaker down.

Therefore, many automakers choose to directly find third parties. The core hardware of the intelligent driving system is the intelligent driving chip. Generally, high-end chips are chosen from Nvidia's Orin and Thor chips, while mid-to-low-end options are selected from domestic chip manufacturers like Horizon Robotics.

BYD and Chery have chosen Horizon's chips for their mid-to-low-end intelligent driving chips.

At the high-level forum of the Hundred People Meeting, Chery's Chairman Yin Tongyue expressed great satisfaction with Horizon's chips, jokingly referring to Horizon's founder Yu Kai, who was also present at the meeting: "Smart heads don't have hair; Horizon's intelligent driving solution is very smart." Yu Kai is bald.

As a Tier 1 supplier in the automotive industry with a century-long history, Bosch Group certainly advocates that automakers seek third parties to solve intelligent driving solutions.

Wu Yongqiao, President of Bosch Intelligent Driving Control Systems in China, predicts that intelligent driving will become a standard feature: "It is not the soul of a car, but a standard feature, just like today's seat belts and airbags. Within three years, vehicles priced between 80,000 and 150,000 will be equipped with mid-level intelligent driving as standard, with computing power of 80-150 TOPS (trillions of operations per second). Three years later, vehicles priced above 150,000 will be equipped with high-level intelligent driving as standard, including highway, urban, and fully automated parking functions." Eddie Wu calculated: "Once intelligent driving becomes standard, the cost of maintaining an AI talent team of one or two thousand people at the OEM will be enormous. Tier 1 suppliers like Bosch, Yuanrong, Momenta, and Huawei can serve seven or eight automotive clients and dozens of models, while the OEM maintaining two thousand people may only handle 10 models, making the overall cost-effectiveness questionable. There’s also the time issue; during the full-stack self-research process, the OEM will need at least three years to close the data loop and fully integrate the toolchain."

Yu Kai, founder of Horizon Robotics, also stated that based on historical data, in 2021, 90% of intelligent driving solutions were self-developed by OEMs, with less than 10% using third-party solutions. However, by 2024, the ratio will shift to 60%:40%. In the future, the ratio will be 20%:80%. This means that most automotive companies' intelligent driving solutions will be provided by third parties.

Range Extender vs. Pure Electric: Who Will Laugh Last?

Achieving 100% autonomous driving on urban roads is still just verbal promotion from automotive companies, and users are not yet able to enjoy it. Currently, the most pressing concern for users is range anxiety.

This anxiety has led to an unexpected shift in the technological route of new energy vehicles in China, as noted by Ouyang Minggao, an academician of the Chinese Academy of Engineering. As the original drafter of China's electric vehicle technology solutions and industrial planning, Ouyang has keenly felt the rapid increase in the sales proportion of hybrid vehicles over the past two years.

He cited a set of data: "In 2024, plug-in hybrids (including range extenders) will account for 40% of new energy vehicles, while pure electric will be 60%. I expect that by 2025, plug-in hybrids (including range extenders) will rise to around 50%, while pure electric will continue to drop to around 50%. This year, the peak of technological transformation for plug-in hybrids should be approaching, and in the coming years, plug-in hybrids (including range extenders) are expected to rise, for example, from 50% to around 60%, while pure electric drops to around 40%."

The main reason for the decline in the sales proportion of pure electric vehicles is that their battery capacity and range have not achieved breakthrough progress technically.

Taking the Tesla Model Y 2025 version as an example, data from Dongche Di shows that its rear-wheel-drive version claims a range of 593 kilometers, but the actual measurement is 395 kilometers, while the long-range all-wheel-drive version claims a range of 719 kilometers, with an actual measurement of 460 kilometers.

However, Academician Ouyang believes that the dawn of breaking through the battery technology bottleneck has appeared: "Domestic major automotive battery companies are choosing high-nickel ternary positive electrodes, sulfide-based composite electrolytes, and silicon-carbon negative electrodes to achieve a target energy density of around 400Wh/kg for automotive batteries, surpassing traditional liquid and solid-liquid hybrid batteries. This technology and product can be industrialized around 2027-2028, and mass production is completely feasible by around 2030. For 500Wh/kg automotive all-solid-state batteries, breakthroughs in lithium negative electrode technology are necessary, which may require the empowerment of AI-based material research and development platforms. The expected industrialization timeline is around 2030-2035, with a five-year uncertainty period." Based on this, Academician Ouyang Minggao believes that after 2030, pure electric vehicles will return to a dominant position with a market share of 70% in overall electric vehicle sales.

However, Academician Yang Yusheng of the Chinese Academy of Engineering, who has always advocated for range-extended technology, believes that the aforementioned timeline is overly optimistic. He stated that it is unnecessary to invest heavily in all-solid-state batteries to equip 1,000 vehicles in three years. This is a waste of human and financial resources.

Currently, the biggest criticism of the plug-in and range-extended technology routes is that they are merely transitional solutions, still relying on fuel and producing carbon dioxide, which does not align with the broader goals of carbon peaking and carbon neutrality.

However, Academician Yang Yusheng also provided a technical solution for range-extended electric vehicles: "Once green liquid fuels represented by methanol and ethanol are abundant, range-extended vehicles will no longer burn oil. At that time, energy-saving, green, safe, long-range, and inexpensive range-extended vehicles will accompany humanity in achieving carbon neutrality."

Nevertheless, this technological route, like all-solid-state batteries, still requires time for verification. For the automotive industry, which undergoes technological iterations every few months, it remains uncertain who will still be at the table when the next generation of technology goes into mass production.

Author of this article: Sun Chunfang, Source: Prism, Original title: "Xiaomi Car Accident Douses the Enthusiasm for Autonomous Driving Nationwide | Prism"

Risk Warning and Disclaimer

The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial conditions, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at one's own risk