On March 18, pulsestacks reported that at the Electric Vehicle 100 Forum held on March 16, Jiyue Auto's CFO, Liu Jining, discussed topics such as Jiyue's competitive differentiation, profit strategy, and the advantages of its pure vision technology solution.
Back in 2021, when Jiyue was founded, the company predicted that the end of 2023 to the beginning of 2024 would be the industry's peak period for advanced intelligent driving technologies. Therefore, Jiyue's models were equipped with advanced intelligent driving capabilities upon their launch. He stated, "Currently, Jiyue's PPA (point-to-point pilot assistance) has covered over 400,000 kilometers of roads nationwide and will achieve nationwide coverage this year. Wherever Baidu Maps reaches, Jiyue's PPA is there."
He believes that in the face of fierce market competition, differentiation is crucial. For the Jueyue automotive robot, it has three main features: natural interaction, autonomous mobility, and self-improvement. Autonomous mobility refers to self-driving capabilities, while natural interaction allows seamless communication between humans and vehicles using natural language. "We are the first to utilize the complete version of the 8295 chip, running a localized speech model, combined with Baidu's AI empowerment in voice technology. Our natural interaction user penetration rate is 98%, meaning almost everyone is accustomed to interacting with the car using voice. On average, this interaction occurs 63 times a day."
Regarding sales, he mentioned that Jueyue has only been delivered for a few months and emphasized the importance of patience and focus. "At this stage, we are concentrating on providing excellent products and services. We believe that sales and long-term financial performance will follow suit. Currently, all our efforts are directed towards serving our initial batch of early adopters and ensuring their satisfaction. Internally, we are more focused on NPS in the short term." To facilitate feedback, Jueyue vehicles are equipped with a one-click voice feedback feature, allowing owners to provide suggestions directly to SIMO via voice commands.
"Another important aspect is self-improvement. We believe that whether it's robots or automotive robots, self-improvement is crucial."
In Liu Jining's view, electrification has disrupted the industry landscape, leading to significant differences compared to five or ten years ago. "I firmly believe that just as electrification has reshaped the industry, intelligence will do the same. There are excellent opportunities for new entrants or players who are well-prepared within the industry."
Regarding profit avenues for automotive companies, he mentioned the importance of scale in achieving financial profitability. "Regardless of the company, reaching a certain scale rapidly and leveraging this scale to generate economies of scale is undoubtedly a fundamental path to profitability from a hardware perspective."
"Now, whether it's autonomous driving or intelligent cabins, they bring many other possibilities for software services to vehicles. Apart from being a commuting tool, cars are also becoming a generation of AI hardware from another perspective. Once software and services are well-established, I believe there will be plenty of opportunities for commercialization."
He also acknowledged that automotive cabins or intelligent driving services are relatively vertically closed ecosystems. If the OEMs can keep up with services, there's an opportunity to gain good commercial returns through excellent service, but this is still in the exploratory phase. "It's too early to say when this will happen or how big the opportunity is. We first focus on delivering excellent user experiences to make users inseparable from this product, and opportunities will naturally arise later on."
Regarding the comparison between pure vision solutions and LiDAR solutions, he mentioned that the LiDAR solution is akin to having two sets of models on the vehicle: one is the vision model, and the other is the model made from LiDAR and point cloud data. Merging these two models achieves intelligent driving. "Our approach focuses on maximizing visual information, using a larger dataset to develop a single large model that incorporates several smaller models. This large model definitely captures more information than the smaller ones."
He indicated that Extreme Speed can obtain sufficient computing power from Baidu. Currently, for various reasons, high-capability computing power is scarce in China, and whether sufficient computing power can be secured also determines the level of vehicle intelligence, including autonomous driving capabilities. Wang Liang, chairman of Baidu's autonomous driving technology committee, has shared that in terms of data volume alone, Extreme Speed's vision with an 8 million pixel camera gathers 160 times more data than point cloud information, obtaining more dense and valuable information. "Our deep mining of data and application of large models undoubtedly have much greater potential than LiDAR."
"We also recognize the unique applications of LiDAR. Beyond the pure vision model itself, we use the OCC (Occupancy) network for the recognition of irregular obstacles, significantly replacing the use of LiDAR. We do not deny the advantages of LiDAR, but we have better algorithms to replace it."
From a cost perspective, automakers face two types of costs: the cost per vehicle and the R&D cost. "For us, our short-term investment is higher than that of using LiDAR, as LiDAR has ready-made solutions for automakers. However, we have invested more in computing power and model consideration, reducing the BOM (Bill of Materials) cost for users. This brings two benefits: one is lowering the entry price for buying cars, allowing users to access advanced intelligent driving at a lower price nationwide."
The second is maintenance cost. "If the LiDAR is damaged, the repair cost will definitely not be just a few thousand CNY, including the subsequent insurance costs, which will be very high. At the same time, LiDAR might also have a lifespan issue."
"Therefore, we believe in using a simpler solution with greater computing power and larger models to achieve better results. If we can do all these, we believe we can also offer more people the enjoyment of advanced intelligent driving at a lower and more favorable price."
Regarding the overseas market, Extreme Speed will participate in the Abu Dhabi Auto Show at the end of April and early May and will showcase some capabilities abroad. "In addition to preparing hardware for overseas markets, such as adapting to local languages and applications in other regions, we also aim to pursue. We still believe in prioritizing the user experience. By striving to make the experience comprehensive, we believe the results will follow."
