A：The company needs to make decisions related to organization (whether to choose an entrepreneurial model or a professional management model), products (the type of products they should build), and technology (the direction in which technology should be headed).

A：The entrepreneurial model and the professional management model differ in management principles and key operating principles. They are tailored to different stages of growth and industry environments. The professional management model relies on relatively stable industry and technology cycles, the enterprise being in a leading position with a stable status, and the founder and founding team being uninvolved or losing motivation. The entrepreneurial model is better suited for rapidly changing environments with an unstable industry and technology cycles, and the company not yet being a leader, with the founder and founding team being actively involved and passionate.

A：The company believes it fits better into the entrepreneurial model due to the rapidly changing environment, the unstable industry, the fact that it is not yet a leader, and because the founder and founding team remain fully motivated and involved in daily work.

A：The four key aspects of the entrepreneurial model that the company focuses on are: 1) fostering deep conversations to increase knowledge and judgment, 2) focusing on user value rather than just short-term delivery results, 3) increasing efficiency while reducing resource occupation, and 4) concentrating on solving key issues rather than creating information asymmetry.

A：The company and its team realized that adapting to the professional management model was not the right fit because it led to them becoming a diminished version of themselves. They recognized the need to embrace an entrepreneurial model that aligns with their strengths and usual practices, especially given their passion for products and rapidly changing technologies.

A：Starting from Q4 this year, the company and the founding team will revert to the entrepreneurial model to embrace the new era and technological challenges.

A：The company needs to make an important choice regarding whether to focus on electric vehicles or smart devices and how to define the product it should build for users. This choice influences whether the focus is on larger space, more range, and cheaper prices for electric vehicles or on features and experiences traditionally found in smartphones and smart tablets within the car environment.

A：The embodied AI in cars is envisioned to have perception like eyes and ears, modeling capability like a brain and nerves, computing power like a heart, and a reshaped hardware form to be a stronger physical presence. It should be able to perform tasks such as driving, picking up and charging cars, opening and closing doors, and providing services like a parent or an assistant to make the user's life more convenient and safer.

A：The speaker suggests that the most valuable embodied AI products in the next decade will be vehicles that are not only automated but also proactive. These products will provide automated and proactive services that change the life of users and are considered to be like the car-shaped robots seen in movies and shows like Transformers. The speaker views this as the biggest challenge and opportunity for entrepreneurs in the new era.

A：The two models for the full stack AI system are language-based and those faced towards the physical world. The language-based model is primarily focused on computational power and is suited for digital tasks. In contrast, an AI system for embodied AI must have a better understanding of the physical world, integrate optimally with hardware and software, operate at higher frequencies, and have computation power that resides on the device itself rather than the cloud. Additionally, it requires tailored chips and the modification of hardware to become embodiment-specific.

A：For embodied AI, significant changes are required in perception models and hardware. These include an increase in the effective range of perception, tailored chips for embodied AI, a strong compiler team for high efficiency, and the integration of 3D vision transformer technology. The goal is to improve the understanding and perception of the physical world, utilize human data more effectively, and enhance the closed-loop response time for autonomous driving systems.

A：By focusing on solving the key issues in every stack on every level of the stack, the next generation of autonomous driving can increase performance by five to tenfold, according to the speaker. This approach looks at the entirety of the control mechanism and seeks to reduce the time from perception to execution, potentially by as much as 550 milliseconds compared to a typical autonomous driving system. This improvement can lead to a significant reduction in accident rates and enhance the overall safety and user experience.

A：The key strategic choices that support the company's future development include focusing on short-term weight, staying on the right path over the long term, and continuing to invest in the right areas, backed by a strong financial foundation.

A：The financial performance results for the third quarter include total revenue of RMB 27.4 billion, a decrease of 36.2% year over year and 9.5% contraction, with RMB 25.9 billion from vehicle sales. The sequential decline in revenue was partially offset by a higher average selling price due to different product mix. The cost of sales was RMB 22.9 billion, down 22.0% year over year and 5.3% over the previous quarter.

A：Gross profit in the third quarter was RMB 4.5 billion, which was 51.6% lower year over year and represented a gross margin of 26.3%. The vehicle margin would have been 19.8% in the third quarter due to the recall cost. Cross market in the third quarter was 16.3%, which would have been 20.4% excluding recall costs.

A：The pay of new vehicles and increased investment in planning the product portfolio and technology, along with expenses from product configuration adjustments, caused the year-over-year increase in research and development expenses.

A：The operating margin in the third quarter was negative 4.3% compared to 8% in the same period last year and 2.7% in the prior quarter. The net loss in the third quarter was RMB 624.4 million, compared to a net income of RMB 2.8 billion in the same period last year and RMB 1.1 billion in the prior quarter.

A：The company expects vehicle deliveries in the fourth quarter of 2025 to be between 100,000 and 110,000 vehicles, with total revenue between RMB 5.65 billion and RMB 29.2 billion. This outlook is based on the company's current and preliminary view on its business situation and market conditions, which are subject to change.

A：The focus within electric drive systems is on efficiency and user experience, with an in-house developed and outsourced manufacturing of silicon carbide power chips, power modules, and motor controllers, as well as establishing a dedicated drive owner factory. The company has developed a full chain in-house development capability from silicon carbide power chips to electric motor assemblies. The battery system focus is on safety, with capabilities around 5 C ultra-charging batteries, full control over chemistry, BMS control modules and algorithms, battery pack layouts, and structural design. The electronic control system aims to provide the best driving experience through in-house developed hardware and software, including full-stack development capabilities for powertrain control, power management, and engine calibration.

A：The company's strategy on the supply front for battery systems is a combined approach of external procurement and in-house development and production. This combined strategy is expected to further strengthen battery safety and improve user experience.

A：The company sincerely apologized to customers who placed orders for the I 6 model and are still waiting for delivery due to supply chain constraints, component planning, and production ramp-up pace. The company's team is working around the clock to accelerate production and expedite the delivery process.

A：The phase-out of the subsidy will lead to a pull-forward effect in sales as customers rush to lock in their purchase before the end of 2025, which is expected to result in a substantial dip in deliveries in Q1 2026. The company is preparing for this with a 'peace of mind' purchase program that covers the purchase tax difference for iix customers who locked in their orders in 2025 but paid delivery in 2026. The company is also ensuring all 2026 models meet new standards for gas and energy consumption to qualify for 2026 incentives. In the longer term, the company will offset policy impacts through technology advancements, such as adopting a 500-800 volts high voltage platform and 5 C ultra-fast batteries, aiming to operate about 4800 supercharging stations by 2026, and leveraging economies of scale and supply chain localization to stabilize pricing and accelerate product iteration.

A：The planned upgrades for the L series in the next year will be a major generational improvement based on user feedback and research, featuring a simplified SKU approach for market coverage and supply chain efficiency. The base model will maintain a strong user experience without compromises, with an upgraded design that strikes a balance between brand identity and user appeal. Core technology will include 5G supercharging for all models, integrated with the existing charging network.

A：The recall was announced and involves lowering the delivery of 2020-25 models to ship replacement units to customers for recall. This action is deemed beneficial for customers and aligns with the company's value proposition.

A：The VLA driver was rolled out to all Ad Max vehicles in September, and the model migration capability allows all Ad Max users, including new i-series and L9 users from 2020, to access the new features. User feedback and data analysis have shown effectiveness and improvements in the L and I series owners' experiences. Future plans include ongoing iteration of the VLA with the release of OTA 8.1 in early December, which will enhance perception capabilities and responsive behavior. An architecture upgrade will also be deployed to strengthen decision-making processes and compatibility with the upcoming 100 chip core system improvements.

A：The in-house developed AI aim for system is a core foundation for intelligent vehicles, featuring an integrated architecture to reduce development time by 20% and costs. The company has an ongoing collaboration with open source OS Ena for a community charter and has established the Technical Steering Committee. A focus is on improving perception, understanding, and response in the physical AI model. The AI inference chip is undergoing large-scale system testing, with commercial development expected to start next year. The next generation VLA-based autonomous driving system is expected to have at least three times the performance to cost ratio of current high-end chips, with continuous performance improvement and cost-effective development as future priorities.