Tesla's Struggle in the Chinese Market Challenges and Prospects
On August 20th, news came that due to difficulties in introducing its most advanced driving assistance functions in the Chinese market, Tesla is slightly behind in the competition with Chinese electric vehicle manufacturers. These local manufacturers have begun to sell vehicles that can cruise automatically on urban streets.
China is Tesla's largest market outside the US. The company hopes local drivers will widely use the software named Full Self-Driving (FSD) to achieve fully automated driving from the starting point to the destination. Although the system can operate independently, the driver still needs to be ready to intervene at any time. Tesla's CEO Elon Musk expected last month that the Chinese regulatory agency would approve these functions that have already been allowed in the US by the end of this year.
But according to people familiar with the official thinking, due to concerns about accidents caused by Tesla's software in the US and data security issues, the Chinese government has not yet approved Tesla's latest driving technology. At the same time, the company has not obtained permission to transfer the data of Chinese car owners to US servers to train its AI model. The unclear expectation of approval prospects prompts Tesla to consider further investing in infrastructure such as computers in China.
Meanwhile, some Chinese automakers have obtained government approval, and the driving assistance functions they offer are not only more advanced than what Tesla can provide in China, but also cheaper. Last month, XPeng launched a driving assistance system nationwide that can cruise on urban streets and respond to traffic signals. XPeng said its system is driven by artificial intelligence.
Other Chinese companies, including Huawei and Li Auto, also provide similar driving assistance functions in hundreds of Chinese cities.
According to people who understand the situation, if Tesla gets the approval of its FSD service in China, it may initially be limited to a few cities such as Shanghai for trial. This is in line with the routine practice of the Chinese regulatory agency: usually requiring local companies to first pilot their services in a limited area.
Currently, the Tesla-assisted driving software available in China costs about $4,500, while Chinese competitors usually charge lower fees or provide the software free of charge to car buyers. Analysts said the price difference makes Tesla's software less attractive to Chinese car owners.
The general manager of Shanghai consulting firm Automotive Foresight, Zhang Yu, when comparing the assisted driving functions that Chinese drivers can currently use, bluntly said: Tesla has lagged behind in China!
Tesla did not respond to the comment request. On its Chinese official website, it pointed out that before providing advanced functions in the US market to Chinese consumers, it needs to accumulate billions of kilometers of driving data and obtain relevant regulatory approvals.
The development of self-driving cars is crucial to Musk and Tesla. Anyone who doesn't believe Tesla will solve the problem of driverless cars should not hold Tesla stocks. He said this in the latest earnings conference call.
Some Chinese automakers also admit that if Tesla's advanced software is approved, it may exceed their own software. Gu Junli, a former technical executive of Tesla and XPeng Motors, said at an AI conference in July that Tesla is leading Chinese companies in driving assistance technology by up to two years.
Slow sales
In the first half of this year, the Chinese market accounted fo r 20% of Tesla's total revenue, but like other US and European automakers, the company's sales in China are declining. According to the data of the China Passenger Car Association, in the first half of this year, Tesla's share in the new energy vehicle market in China dropped from 9.5% in the same period last year to 6.8%.
Tesla said its FSD driving software version 12 adopts artificial intelligence and learning tools that imitate the neural network of the human brain, and is frequently updated with new sub-versions. Since the beginning of this year, US drivers have been able to use this version, and in most cases the vehicle can drive by itself, including on urban streets, without the need for manual steering wheel rotation or accelerator pedal depression, but Tesla still warns that the driver must be ready to take over control at any time. In contrast, the Chinese regulatory agency has not yet allowed vehicle companies to provide systems that allow hands-free driving.
Tesla has adopted a brand new software architecture based on artificial intelligence, abandoning the complex rule programming that previously tried to preset all possible situations that the car might encounter. The company announced in March that its latest system has been trained through millions of video clips, replacing hundreds of thousands of lines of computer code.
Meanwhile, Chinese automakers are also actively researching and training their own artificial intelligence models. Similar competitions of electric vehicles have helped Chinese automakers move to the global forefront.
UBS analyst Gong Min pointed out: The beauty of this market is that because of the diversity of participants and fierce competition, it promotes more trials and errors, which accelerates the progress of the entire industry.
Green light
During his visit to Beijing in April this year, Musk won initial support for the Tesla FSD promotion plan, significantly boosting the company's share price and paving the way for Tesla to start testing vehicles and collecting the data required for regulatory approval. In May, the relevant Shanghai department allowed Tesla to transfer the data of Chinese test vehicles to the US to a limited extent, but the video data must not contain human faces or license plate numbers.
According to people familiar with the matter, Tesla is currently testing a small fleet of vehicles with the latest driving software in Shanghai and is preparing for trials in Hangzhou and Beijing. In addition, Tesla is also actively seeking permission from the Chinese side, hoping to transfer the data of ordinary road vehicles (not just test vehicles) to the US, and has discussed methods on how to filter sensitive data. However, the future is still unclear.
In the face of challenges, Tesla considered an alternative plan - to establish a data center in China to train driving software under Chinese road conditions. However, this plan is not simple. Tesla is studying whether it can purchase the most advanced Nvidia chips to power this artificial intelligence effort, but the US currently restricts the export of these chips to China.
In the US, federal regulatory agencies have investigated hundreds of accidents involving Tesla vehicles using earlier driving assistance software, including some fatal accidents. In December last year, in order to increase driving vigilance, Tesla recalled more than 2 million vehicles and added more alerts and control measures. People familiar with the matter said that these accident reports have had a significant impact on the review of Tesla technology by the Chinese regulatory agency.
The artificial intelligence driving system of Tesla has attracted additional attention at the Chinese regulatory level, mainly because it is worried that the system may not be able to effectively deal with the complex traffic environment in China without sufficient local road training. The unique driving conditions in China, such as the changeable road conditi ons in small cities and the dense flow of bicycles and pedestrians, are different from the training conditions in the US, which has aggravated the concerns of the regulatory authorities.
XPeng Motors founder He Xiaopeng experienced Tesla's latest functions in California in June this year. Although he said he was impressed, he also pointed out: The road conditions in Chinese cities are about ten times more complicated than those in the US.
IT Home note: The news is from The Wall Street Journal.