Preface:

On February 6, Geely announced the successful integration of its self-developed Xingrui large model with the DeepSeek-R1 large model, positioning itself as the automotive industry's pioneer in leveraging DeepSeek technology, primarily for intelligent cabins. Other prominent enterprises such as Dongfeng, IM Motors, Changan, Leapmotor, and SAIC have also made significant strides in integrating DeepSeek into their intelligent systems. In contrast, BMW will not integrate DeepSeek into its new models until the latter half of this year, appearing to lag behind.

Author | Fang Wensan

Image Source | Network

BMW's Urgent Need for Intelligent Transformation

BMW has long been a frontrunner in the automotive industry with its manufacturing prowess and brand image, yet its intelligence development has lagged somewhat. When compared to emerging domestic automakers, BMW's in-car system falls short in user experience, a significant disadvantage in today's fiercely competitive market.

BMW's dominance in the Chinese market was once unparalleled, with sales reaching as high as 64% in 2020, but this figure has now dwindled to around 32%. In 2024, BMW's sales in China declined by 13.4% year-on-year, while sales of models equipped with Huawei's HarmonyOS surged by 27% during the same period.

Behind this sales decline lies the rapid evolution of Chinese consumers' demand for intelligent experiences. In response to this shift, BMW must reassess its innovation strategy.

Intelligence has emerged as the predominant trend in the automotive industry. Consumers now prioritize features such as intelligent interaction and autonomous driving over traditional performance and aesthetics.

BMW's intelligence development has not kept pace, with a noticeable gap between its in-car system's functionality and user experience compared to domestic competitors. This disparity is not solely due to technical limitations but also reflects a lack of sensitivity to market changes and the speed of adaptation to new trends. Consequently, accelerating its digital transformation has become a pressing issue for BMW.

For global automotive giants like BMW, technological advantages once secured their dominance in the market. However, with the rapid advancement of China's automotive industry and technological prowess, these traditional powerhouses must reconsider their strategies.

BMW China's Integration with DeepSeek: Bolstering the 360-degree AI Strategy

Shortly after BMW unveiled its 360-degree full-chain AI strategy in March this year, leading Chinese tech enterprises such as Alibaba, Huawei, ByteDance, and DeepSeek became strategic partners. At the Shanghai Auto Show, BMW showcased its [Car Expert] and [Travel Companion] intelligent systems, developed in collaboration with Alibaba and led by BMW's local R&D team.

On April 18, BMW announced its strategic partnership with Volcano Engine, a ByteDance subsidiary, focusing on innovative AI applications in automotive marketing services to enhance user car-buying experiences and dealer operation efficiency through AI-driven content recommendations.

Building on its AI large language model collaboration with Alibaba, BMW further advanced its 360-degree AI strategy. On April 27, BMW Group announced that starting from the third quarter of 2025, DeepSeek functions would be introduced to multiple new vehicles equipped with the ninth-generation BMW operating system, including the BMW 5 Series Long Wheelbase, the fully electric BMW i5, and the all-new BMW X3 Long Wheelbase.

Following previous collaborations with Alibaba and ByteDance to build an intelligent marketing system, DeepSeek will now extend the intelligent experience into the cockpit. These initiatives underscore the pivotal role of the Chinese market in BMW's next-generation strategy.

To date, BMW has deployed over 600 AI applications globally, with its first AI application in China dating back to 2018. These applications enable the BMW new-generation intelligent cabin to handle complex multi-command queries and provide suggestions that align with the BMW brand.

Through its partnership with DeepSeek, BMW aims to maintain its mechanical performance edge while addressing localization deficiencies in the intelligent cabin.

Since its inception in March 2025, BMW's 360-degree full-chain AI strategy has encompassed 18 core scenarios, ranging from supply chain optimization to user lifecycle services.

Embracing Domestic AI: A Shift in Industry Discourse Power

Engineers at the Shanghai Auto Show revealed that "the new system's Chinese dialect recognition accuracy has reached 85%, and we are currently addressing accent differences in the Jiangsu, Zhejiang, and Shanghai regions." This cultural adaptability poses a significant challenge for foreign brands seeking independent development.

DeepSeek's R1 inference model and V3 large language model are tailored for the Chinese context, offering robust logical reasoning capabilities and widespread adoption by mainstream Chinese automakers like BYD, Geely, and Great Wall Motors.

DeepSeek's cost-effective models have directly benefited BMW by increasing its gross profit margin by 2 to 3 percentage points.

More importantly, this collaboration could signal a new paradigm for international automakers. Traditionally, foreign headquarters handle R&D, while the Chinese region focuses on localization. However, BMW is now feeding AI solutions from the Chinese market back into its global R&D system.

Recently, the Munich headquarters has sent numerous engineers to China to learn from local experiences, a phenomenon almost unthinkable five years ago. This transformation reflects the harsh realities of market competition, involving a delicate balance between traditional automotive manufacturing concepts and digital-era thinking, global uniformity versus local characteristics, and brand premium versus technology popularization.

BMW China's decision to integrate DeepSeek is not a compromise but a proactive strategic shift. Identifying problems is crucial, but proactively addressing them is even more so.

DeepSeek + Automobile: Accelerating the Era of [Intelligent Driving Equalization]

DeepSeek excels in advanced algorithms, data processing, optimization, continuous learning, and user experience enhancement. As an open-source foundation model, DeepSeek is poised to accelerate intelligent driving training, reduce costs, and become an indispensable tool in this domain.

During the automotive R&D phase, DeepSeek aids engineers in virtual simulation testing, expediting new model development and cost reduction. In intelligent driving system development, traditional perception models rely heavily on cloud computing power and data training costs, requiring vehicles to be equipped with high-powered computing platforms.

DeepSeek's impact on the intelligent cabin is more direct: The DeepSeek-R1's strengths in interactive experience, openness, and cost fully meet automotive intelligent cabin needs. Its large language model-based interoperability can be seamlessly integrated into the intelligent cabin voice system, appealing to automakers.

In the cloud, DeepSeek reduces data annotation requirements and enhances model capabilities through distillation, lowering vehicle-side computing resource demands. By distilling technology to create a small, high-performance model, if DeepSeek-R1's innovative algorithm architecture is applied to intelligent driving system development, it could significantly improve training efficiency and reduce costs.

Furthermore, DeepSeek's training mode of cold start data + unsupervised reinforcement learning offers new perspectives for intelligent driving system training. Its application could lower annotation costs and enhance model generalization capabilities. Combined with DeepSeek's multimodal capabilities to simulate long-tail training scenarios, it could propel the technical maturity of intelligent driving systems.

While DeepSeek's impact on the intelligent driving industry's competitive landscape is modest, its significance lies in accelerating the era of [intelligent driving equalization] by lowering the barriers to entry for advanced intelligent driving technologies and capital.

Conclusion:

In an increasingly competitive environment, respecting and fully leveraging the unique resources of the Chinese market has become crucial for international automakers' success in China. Establishing deep partnerships with leading domestic technology companies is an effective strategy to achieve this goal.

References: DianChang: "BMW Speeds Up in China: From Local Production to [Home in China] Deep Integration into the Intelligent Ecosystem", JiChe ZhiZao: "New Generation: Can It Open a New Era for BMW?", JiangHanShiYe: "BMW China Announces Integration with DeepSeek, Is BMW Compromising?", GuangDianLianJie: "Just Now! BMW Announces Cooperation with DeepSeek"