Throughout the interview, Li Xiang also provided the future development direction of Ideal Autos, and he believes that Ideal Autos will definitely continue to focus on AI, ultimately achieving AGI (Artificial General Intelligence) through the platform of Ideal Autos.

Regarding future product layout, Li Xiang directly denied the Robotaxi product, stating that Ideal Autos will not focus on this area in the future, but will absolutely create humanoid robots, although this is not happening right now.

By 2030, there is a 50% probability that Ideal Autos will produce a very interesting supercar that will be equipped with AI.

Throughout the interview, Li Xiang mentioned the direction of AI multiple times, indicating that the next stage for Ideal Autos will focus primarily on AI, which is a consensus among many auto manufacturers. Previously, Xiaopeng Motors had also discussed AI several times at their Technology Day.

Additionally, a quite interesting point is that the background of this interview with Li Xiang was during the daytime, which should indicate that it was a pre-recorded session. In the next two days, Li Xiang will continue to release two more interviews, introducing more aspects of Li Auto.

01. Li Xiang aims to be the CEO of an AI company, divided into three phases.

Regarding the future product planning of Li Auto, Li Xiang stated that Li Auto now aims to be an AI company, and he himself intends to transition from a car company CEO to an AI company CEO.

Li Xiang believes that what Li Auto should focus on is not the intelligence of Autos, but the automation of AI in Autos. The logo of Li Auto has never included the word 'Autos,' and even now, the operational company is still called 'Peking Che He Jia Information Technology Co., Ltd.'

However, Li Xiang also mentioned that Hardware is very important. He describes Li Auto as 'connecting the physical world and the digital world,' enabling it to become a leading AI enterprise.

Li Xiang mentioned that in the era of the Penghua CSI Mobile Internet Index Fund(LOF)-A, he chose a very vertical field; although one may do well, in some ways, one might miss the forest for a single tree. Therefore, when choosing to embark on his third entrepreneurial journey, Li Xiang stated that he wants to choose a forest and aim to be the largest in that forest, regardless of the difficulties that may arise, definitely not just aiming for a single tree.

Regarding how to become an AI enterprise, Li Xiang stated that Li Auto needs to develop its own foundational model, as AI represents everything for the future of Li Auto.

As for how product forms will evolve with advancements in AI capabilities, Li Xiang described the realization of AI AGI (Artificial General Intelligence) and its final application in three phases.

Li Xiang believes in three stages of AGI (Artificial General Intelligence).

The first stage, which Li Xiang refers to as "enhancing my capabilities."

Li Xiang explains that AI serves as an aid to the user, but the final decision-making authority lies with the user.

Li Xiang believes that Level 3 autonomous driving still requires supervision in the car to provide final responsibility; the core reason is that the capabilities at the first stage are still insufficient, and the user is responsible, hence it is referred to as enhancing the user's capabilities, but it indeed makes the user very convenient and significantly improves the user's efficiency.

The second stage, which Li Xiang refers to as "becoming my assistant."

Li Xiang explains that the user only needs to assign tasks to the AI, and even continuous tasks can be completed independently by the AI, which will take responsibility for the results.

Li Xiang provides an example where the user can tell a Level 4 car to help pick up a child; the user does not need to be in the car, and it can go to the school to pick up the child, performing facial recognition to unlock the door and let the child get in.

Li Xiang believes that a good stage is that AI will become widely applied, which is the real phase of iPhone 4. For auto companies, only achieving L4 is the true iPhone 4 phase, but Li Xiang thinks today is not it.

The third stage is what Li Xiang considers the ultimate stage of AI AGI, which he calls 'Silicon-based Family'.

Li Xiang explains that in this stage, users no longer need to give any instructions to AI, nor do they need to assign tasks to AI; AI becomes a member of the family and even an important organizer within the family. AI not only understands the user but also the user's children and friends, possibly even better than the user does.

Li Xiang believes that by this stage, AI will actively engage in many tasks and can autonomously measure and help manage the household. When AI AGI develops to the third stage, a key point Li Xiang considers important is that the user's memories will also be preserved by AI. The user's physical existence may fade, but their memories will become part of the AI.

02. Completely focus on robots, not interested in Robotaxi.

Regarding the competition in the AI AGI industry, Li Xiang believes that today is still a very preliminary stage. Everyone is still working to obtain the L3 ticket for AGI (General AI) and the L4 ticket for autonomous driving. Since the Ideal Company is doing both in these areas, Li Xiang sees this as an opportunity that must be pursued.

Regarding large models, Li Xiang believes that base models at a certain point will inevitably evolve into VLA (Vision Language Action Model).

Li Xiang believes in the importance of the VLA (Vision Language Action Model).

Li Xiang explained that language models also need to understand the three-dimensional world through language and cognition, and the three-dimensional world is not only images; images cannot restore the real physical world, and it is necessary to truly understand the world using Diffusion models.

Li Xiang believes this is the same for autonomous driving; to reach L4, autonomous driving must have a strong cognitive ability to effectively understand the world when changes occur, rather than just relying on the compressed memories behind an end-to-end model.

Therefore, for the Li Auto team, Li Xiang requires that within China, the foundational model of large language models must be among the top three in the industry in the coming years. Regarding this requirement, Li Xiang is willing to invest in whatever training computing power is necessary.

Regarding Robotaxi, Li Xiang stated he does not want to pursue it. Li Xiang believes that once L4 autonomous driving is achieved, family cars will become cheaper and costs will decrease, leading to more people willing to own a car.

Li Xiang believes that when a space becomes better, more efficient, and offers better experiences, users should have ownership of this space; the ultimate mobile home is actually L4, and a happy home is what Li Xiang refers to as a 'silicon-based family.'

Regarding whether Li Auto will develop humanoid robots, Li Xiang stated the probability is definitely 100%, but the timing is not now. Li Xiang believes L4 autonomous driving is the simplest form of robotics, and if L4 autonomous driving cannot be solved, it will be impossible to tackle more complex humanoid robots.

In addition, Li Xiang also responded to the recent hot topic of why to buy a Ferrari; he believes experience is crucial, as only through personal experience can it transform into one's own cognition and ability.

Li Xiang also proposed that by 2030, Ideal Autos would have a 50% chance of making a very interesting supercar, but it would definitely be an AI sports car.

03. Conclusion: Li Xiang fully bets on AI.

After 'disappearing' from the public eye for nine months, Li Xiang has returned to the spotlight.

However, this time, Li Xiang did not launch any products, but rather spoke extensively about AI. He views AI as 'everything in the future,' believing that the emergence of large models will bring about fundamental changes for humanity, and he also discussed the latest developments in Ideal Autos' intelligent driving and the use of AI technology in the Ideal classmate app.

So what will an AI Ideal Autos look like in the future? Let's wait and see.