Haitong International: Investors are optimistic about the future widespread adoption of smart driving chips, and the shift of third-party chip suppliers to Tier1 is a crucial factor for competitiveness in the next stage.

Sealand International released a research report stating that according to media reports, BYD (01211) has adopted the black sesame intelligent (02533) automotive-grade autonomous driving computing chip, and the model is BYD's brand of DDMG. The brokerage believes that the DDMG's point of black sesame intelligent is old news, and the recent market enthusiasm around AI and smart driving has really pushed up the stock price.

The report said that from the popularity of DeepSeek, to the frequent exposure of humanoid robot products in recent days, to BYD's upcoming "Intelligent Driving Equity" strategy, the capital market is looking for high-quality AI and hardware integration. In addition, the recent valuation explosion of Horizon Robotics also triggered the re-pricing of domestic intelligent driving chips, and black sesame intelligent as a company in the same track naturally ushered in a rebound. This round of rise is driven by sentiment and also reflects the change in market expectations, showing investors' optimism about the future popularization of intelligent driving chips.

The brokerage believes that the priority of algorithm iteration is still the first, and the integration of software and hardware is the key for chip manufacturers to break through. As the sales leader in the mass market, BYD's "Intelligent Driving Equity" strategy this year may fully open the accelerated penetration of high-end intelligent driving in the industry, and the L2+ function is expected to cross the turning point and enter the mainstream market ahead of schedule, with a large growth certainty of the 100TOPS+ high-performance platform. Although the market expansion is beneficial to upstream, it believes that the intelligent driving industry has certain specialties at present: the algorithm end is parallel with AGI, and the multi-segment/Rule-based architecture is fully transformed into one-segment/Learn-based architecture, and the downstream resource investment will still focus on upper-layer algorithms. Therefore, for downstream self-developing carmakers/algorithms suppliers, the NVIDIA platform with strong CUDA ecosystem, larger computing power support, and higher development and deployment efficiency is still the first choice; for Tesla, Huawei and other self-developed hardware systems, the self-owned hardware know-how has obvious advantages in the utilization of computing power performance, software and hardware development cooperation, and cost optimization, and it is difficult for third-party chip suppliers to break the barriers.

The brokerage believes that the development route of intelligent driving is still steep, and the leading position in a certain stage is difficult to form absolute barriers, and staying on the table is as valuable as leading the industry. As a third-party chip supplier, it is crucial to shift from providing partial software support to hardware Tier2 to Tier1 or maintaining its competitiveness in the next stage.