2024.07.10 12:17
I'm PortAI, I can summarize articles.

Tesla is causing a huge sensation

The next disruptive node

Author | Chai Xuchen

Editor | Zhou Zhiyu

Tesla, which has been frequently absent from auto shows, is making a strong presence at AI exhibitions.

At the recent World Artificial Intelligence Conference held in Shanghai, Tesla became one of the most eye-catching booths. From the Optimus humanoid robot to the Cybertruck equipped with FSD, all are telling Elon Musk's AI story.

Tesla is also stirring up the AI ecosystem with these innovations, seeking breakthroughs for the application of AI. Among them, Tesla's most close-to-production smart driving system FSD has begun to influence the automotive industry landscape.

At the end of last year, the FSD V12 version equipped with an "end-to-end" AI large model began testing in the United States, creating a wave of smart driving trends with human-like driving performance.

The heat has spread to China, bringing hope of breaking the cycle to players who are struggling. Not only new forces like Nio, but also traditional automakers such as BYD, Great Wall, and SAIC are aggressively advancing in smart driving; tech companies like Baidu and Didi are also accelerating the iteration of smart driving technology.

Tesla, the giant whale stirring up China's new energy vehicle progress, is once again causing a storm in the AI field, bringing about a second round of industry reshuffle.


In the past 10 trading days, Tesla has achieved ten consecutive increases, with the stock price rising by 43.68% to above $260, reversing its performance below the market average earlier this year.

Investors' increasing expectations for Tesla's FSD revenue generation and anticipation of the Robotaxi set to be released in August have been the main factors driving Tesla's current uptrend.

The timing of FSD landing in China is also approaching. In early June, a Tesla employee "accidentally" leaked during a live broadcast that the internal test version of FSD V12 was likely being tested in China. At the end of June, Tesla's Vice President Zhu Xiaotong, when asked about the timing of FSD landing in Shanghai, responded with "soon."

In fact, two months ago, when Musk suddenly arrived in Beijing, the process of FSD entering China seemed to have been fast-tracked. Soon after, Tesla was included in the China Association of Automobile Manufacturers' data compliance list; Baidu granted mapping qualifications and provided lane-level navigation systems.

As information compliance issues are gradually resolved, FSD, which had already been dominating in North America at the end of last year, finally has the opportunity to compete in China. The "secret weapon" of FSD, the "end-to-end" technology powered by AI large models, is the core of this year's competition in the smart driving field.

Industry insiders point out that the end-to-end architecture of FSD V12 integrates "perception" and "decision-making" through AI driving models. It is equivalent to the "ChatGPT" of autonomous driving.

Musk emphasizes that FSD V12 does not have any manually set code on how to handle various driving situations; just "feed" it videos, and AI can learn to drive on its own. This has reduced the code by 300,000 lines compared to the past, greatly improving the system's simplicity and operational efficiency Domestic competitors with keen senses have already taken action.

As one of the top players in the smart driving experience in China, XPeng announced this year that it will invest 3.5 billion and add 4,000 employees to aggressively advance AI and smart driving; Huawei launched ADS 3.0, achieving modelization of decision planning; Li Auto has regarded intelligent driving as the "number one project" since last year; even Wang Chuanfu, who was once "ambiguous" about autonomous driving, has now recruited 4,000 engineers and plans to invest over 100 billion in research and development.

This is just a glimpse. Starting this year, whether it is traditional automakers undergoing a "second entrepreneurship" or joint ventures like Toyota, Volkswagen, and General Motors, they are all quickly developing their own technologies or seeking technical assistance to fill the gap in smart driving. They are also promoting the concept of "oil-electric intelligence" on traditional vehicles to capture consumers' minds.

On one side are global electric vehicle giants, on the other side are the watchful domestic players. The drama of "masters competing" is about to unfold, and even the previously indifferent capital is showing renewed enthusiasm.

In May of this year, the UK startup Wayve announced a $1.05 billion financing, which, in the context of the global smart driving financing market entering a downturn in the first half of 2021, is seen as an endorsement of the end-to-end smart driving route by investors.

The smart driving frenzy led by FSD has brought new development opportunities to new energy players once again.


By 2024, the competition in the domestic automotive industry has reached an unprecedented level of intensity.

Former new forces like Nio are no longer at the top of sales, dropping from a peak of 40-50,000 units; Li Auto, which surged last year, was brought back to reality by the MEGA incident, with this year's sales target of 800,000 units turning into a mirage; XPeng's sales in the first five months did not exceed 10,000 units; "big brother" BYD's year-on-year growth rate has also dropped to 20%.

Players lagging behind are eager to turn the tide; cross-border players like Huawei and Xiaomi are using disruptive marketing methods to compete for discourse and pricing power; while the giants at the table are more interested in consolidating their positions and not losing the cities they already have.

In order to defend or break the situation, besides piling up configurations, everyone can only push prices to new lows, and industry competition seems to be heading towards disorder. Zeng Qinghong, Chairman of GAC Group, publicly shouted, "This is not the way to go. Losses are the biggest harm to the automotive industry."

Elon Musk, who stirred up the smart driving track with FSDV12, has unearthed new demands and opportunities to disrupt the game for the market.

Open Source Securities predicts that by 2025, the penetration rate of L3 and L4 will quickly exceed 10%; by 2027, it will reach 30%, and by 2030, it will reach 70-80%.

This is similar to the process of Tesla's localization in China five years ago. At that time, after the localization of Model 3 and Model Y, the penetration rate of new energy vehicles in China quickly surpassed 4% to 50%, and Tesla's FSD is also expected to replicate this path in the field of smart driving.

The Chinese market also believes that Tesla, as a "giant" in new energy vehicles, will transform into a "giant" in smart driving once again and reshuffle the automotive market The warm-up matches are already in full swing. XPeng and Aito are focusing on the intelligent driving label, Xiaomi Motors has iterated the intelligent parking version, and Musk personally sent an email, requesting that the sales staff must allow the owners to experience FSD before delivery, and has cut the FSD subscription price by one-third.

However, under the new competition system, it is still difficult to determine who will take the lead.

In the eyes of industry insiders, advanced intelligent driving has not yet reached its final form. Musk stated that each version of FSDV12 is improving at an exponential rate, and will eventually achieve unsupervised fully autonomous driving; XPeng, Huawei, and others are also aggressively pushing forward, adopting an "end-to-end" approach, without strictly adhering to pure visual or lidar solutions.

How will the market evolve next? Chen Tao Capital has pointed out several potential challenges and opportunities in the current intelligent driving industry, which may become variables for future differentiation among players.

Their released "End-to-End Autonomous Driving Industry Research Report" indicates that in addition to technical route divergences, end-to-end intelligent driving has unprecedented requirements for training data volume and computing power.

This year, Musk, who is "obsessed" with AI, is planning to increase the training computing power for Tesla. The plan is to increase the number of Nvidia H100 chips to 85,000 within the year, reaching the same level as Google and Amazon. Previously, Tesla also deployed a larger-scale A100 GPU training cluster, leading significantly in actual training computing power investment in the intelligent driving industry.

In addition, Musk recently revealed information about the next-generation in-vehicle computing platform HW5.0, with computing power 10 times higher than the current HW4.0, set to be launched in the second half of 2025. This may lead FSD to show a trend of "overwhelming dominance."

However, FSD, which is thriving in the North American market, may face challenges when entering the actual roads in China; on the other hand, Chinese consumers are still relatively resistant to "subscription payment," and pricing may become a potential obstacle. This provides domestic players with some opportunities to "counterattack."

The direction of this new round of competition is still difficult to predict, but the participants have long been preparing. Perhaps according to industry forecasts, in one or two years, when intelligent driving can reach an "iPhone moment," the market landscape will undergo a complete transformation