![]() |
市場調查報告書
商品編碼
1806656
中國供應商的智慧駕駛海外展開(2025年)Research Report on Chinese Suppliers' Overseas Layout of Intelligent Driving, 2025 |
智慧駕駛海外佈局研究:海外佈局面臨多重課題,與海外供應商輕資產合作似乎是目前的最佳方案。
預計2026年將是中國智慧駕駛供應商海外佈局的元年。從市場規模來看,中國高端智慧駕駛市場最終可能被限制在五家以下。這直接加速了產業洗牌和資本整合的步伐。二級智慧駕駛供應商的獨立生存將更加困難。即將被四維圖新收購的PhiGent Robotics就是一個典型的例子。
雖然一些主要的智慧駕駛供應商已經獲得海外汽車製造商的訂單,但通常要到2027年才能開始裝機。這對於面臨現金流短缺的二級和三級供應商來說,幫助不大。
中國智慧駕駛市場競爭激烈,促使中國供應商積極尋求海外發展機會。然而,智能駕駛海外擴張面臨許多課題。例如,海外擴張需要投入大量資源和資金來解決資料合規性和閉環系統問題、智慧駕駛體驗的在地化發展,以及全面整合研發流程和安全文化。
課題一:在地化使用者體驗與信任障礙
中國企業在智慧駕駛海外擴張時,首先要面對地區間駕駛文化的巨大差異。例如,歐洲司機雖然速度很快,但嚴格遵守規則,包括 "左側超車" 規則。
其次,不同地區對ADAS功能的重視程度也有差異。例如,在東南亞,道路上有許多獨特的車輛,例如摩托車和嘟嘟車。智慧駕駛系統必須能夠探測到這些快速移動的小目標,並且能夠容忍相對缺乏監管的交通系統,同時又不會過於頻繁地發出警告。雖然日韓駕駛通常遵守規則,但他們的道路較為狹窄,這要求智慧駕駛系統能夠適應狹窄道路,並採取保守的駕駛風格。
在激烈的市場競爭中,中國本土的智慧駕駛產品通常傾向於透過功能堆砌來吸引消費者。然而,這種策略在海外市場可能行不通。
課題二:智慧駕駛海外擴張的資料合規與閉環
要在海外市場推廣和使用智慧駕駛系統,資料的收集和儲存首先必須符合當地的法規。各國和地區都有嚴格的資料保護法律,例如歐盟的 "一般資料保護規範" (GDPR)。這些法律對 "個人資料" 的定義以及資料的收集、處理和儲存方式提出了嚴格的要求。
智慧駕駛公司在收集道路數據時,不可避免地會收集車牌、人臉圖像等敏感訊息,因此數據脫敏至關重要。此外,大多數法規要求資料必須本地存儲,迫使公司在海外建造或租賃資料中心,這兩者都會增加營運成本。
此外,資料傳輸和處理也受到嚴格限制。智慧駕駛技術的迭代和演算法的最佳化高度依賴於大量的真實數據。中國的研發團隊必須將海外收集的駕駛資料傳回國內進行分析和模型訓練。然而,跨境資料傳輸受到更嚴格的監管。出於資料主權和國家安全的考慮,許多國家明確禁止或嚴格限制敏感資料的出境。這迫使智慧駕駛公司在海外建立本地資料處理和演算法訓練團隊。
本報告對全球汽車產業進行了調查和分析,並提供了中國供應商在海外智慧駕駛擴張的信息,包括區域法規和發展路徑。
Research on Overseas Layout of Intelligent Driving: There Are Multiple Challenges in Overseas Layout, and Light-Asset Cooperation with Foreign Suppliers Emerges as the Optimal Solution at Present
2026 is expected to be the first year for Chinese intelligent driving suppliers to go overseas. In terms of market capacity, China's high-level intelligent driving market may eventually accommodate no more than five suppliers. This directly speeds up the pace of elimination and capital integration in the industry. It is even more difficult for second tier intelligent driving suppliers to survive independently. A typical example is PhiGent Robotics, which is to be acquired by NavInfo.
Some leading intelligent driving suppliers have secured orders from overseas automakers, but the implementation will basically not start until after 2027. This does not help much for second- and third-tier suppliers that are facing cash flow shortages.
The Chinese intelligent driving market is highly competitive, so Chinese suppliers are also seeking overseas market opportunities. However, there are many challenges in overseas layout of intelligent driving. For instance, in the overseas layout, issues related to data compliance and closed loop, localized development of intelligent driving experience and trust barriers, and comprehensive alignment between R&D process and safety culture all require significant investment of resources and funds to address.
Challenge 1: Localization of User Experience and Trust Barriers
When Chinese companies lay out intelligent driving overseas, they are first confronted with the huge differences in driving cultures across regions. For example, European drivers drive at high speeds but strictly abide by rules, and regulations such as "overtaking on the left" must be rigorously followed.
Secondly, the level of emphasis on ADAS functions varies from region to region. In Southeast Asia, for example, there are a large number of motorcycles and unique vehicles like tuk-tuks on roads. Intelligent driving systems need to detect these small, fast-moving targets and tolerate somewhat unordered and unregulated traffic systems to avoid much too frequent warnings. Drivers in Japan and South Korea generally follow rules well, but due to the narrow roads, intelligent driving systems need to be proficient in narrow-road driving and adopt a conservative driving style.
In fierce market competition, domestic intelligent driving products in China generally tend to attract consumers by stacking functions. However, this strategy may not work in overseas markets.
Challenge 2: Data Compliance and Closed Loop in Overseas Layout of Intelligent Driving
To promote and use intelligent driving systems in overseas markets, first of all, data collection and storage must comply with local regulations. Different countries and regions have their own strict data protection laws, such as the General Data Protection Regulation (GDPR) in the European Union. These laws set high requirements for the definition of "personal data" and how to collect, process, and store the data.
When intelligent driving companies collect data on roads, it is inevitable to collect sensitive information such as license plates and facial images, so data desensitization has become a necessary step. Moreover, most regulations require data to be stored locally, which means companies have to build or rent data centers overseas. All of which drives up operating costs.
Furthermore, data transmission and processing are also severely restricted. The iteration of intelligent driving technology and the optimization of algorithms are highly dependent on massive amounts of real-world data. Chinese R&D teams need to transmit the driving data collected overseas back to China for analysis and model training. However, cross-border data transmission is supervised more strictly. Many countries, out of consideration for data sovereignty and national security, explicitly prohibit or strictly restrict the export of key data. This forces intelligent driving companies to establish local data processing and algorithm training teams overseas.
Challenge 3: Comprehensive Alignment between R&D Process and Safety Culture
Conflict in R&D Process
Status Quo in China: During R&D, requirements often change, development and testing are often carried out simultaneously, and document flow is not very complete. Domestic companies pursue the speed of function launch to seize the market as soon as possible.
International Requirements: Companies must strictly follow standardized development processes and comply with standards such as ASPICE and ISO 26262. Detailed documents, strict reviews, and traceable evidence are required for every step from requirement analysis, system design, coding to test verification.
Core Conflict: International customers cannot accept "black-box" software with unclear requirement traceability, incomplete test documents, and inability to prove the functional safety level. What they buy is not just the function, but a complete set of auditable and reliable development processes.
Conflict in Safety Culture
Status Quo in China: Safety assessment, testing, and verification are often conducted in the later stage of product function development, with the goal of ensuring that the product meets the regulatory certifications and standards required for launch on market.
International Requirements: Safety is not a single link but a culture. Safety considerations are integrated into every link from the initial product design, requirement analysis, coding to test verification, emphasizing proactive risk prediction.
Core Conflict: For example, if a Chinese company discovers a defect in a corner case, and the probability of this case occurring is extremely low, it may temporarily set it aside to avoid affecting delivery. However, European automakers will regard this as a major process loophole, and will take measures and spend more time solving it, and may even suspend the project.
For Chinese second and third-tier intelligent driving suppliers that are already facing cash flow shortages, each challenge in overseas layout may become a financial burden that crushes them. Therefore, cooperation with foreign suppliers allows for "light-asset" operation.
Foreign supplier partners, with their deep understanding of local traffic regulations, access standards (such as UNECE regulations in the EU), and certification processes, can guide targeted functional adjustments to products and collaborate with local authoritative testing institutions (TUV SUD, TUV Rheinland, etc.) to assist in handling cumbersome application and testing procedures, thereby greatly shortening the certification cycle. The mainstream HD map providers in Europe are HERE and TomTom.
Yaxon Connect has forged a close partnership with HERE Technologies since 2023. The two parties have jointly provided Chinese automakers with one-stop overseas solutions integrating "compliant access + technology upgrade + localized adaptation" in three major fields: Intelligent Speed Assistance (ISA), ADAS HD maps, and overseas navigation.
In 2023, Yaxon Connect developed and adapted its self-developed ISA map engine to help export automakers obtain EU ISA system certification. In 2024, the two parties jointly launched an e-Horizon (EHP) and Map Engine based on the ADASIS V2 protocol. This solution has been applied to the Predictive Adaptive Cruise Control (PACC) system of leading automakers, which can effectively reduce energy consumption by 8%-12%.
Up to now, the cooperation achievements of the two parties have been successfully implemented in multiple leading commercial bus and truck OEMs in China, and the products have been exported to such markets as the EU, South Africa, Australia, Mexico, and South America.
After front-end data collection, the raw data is initially cleaned and structured via edge computing devices deployed on test fleets, and then stored in local data centers of international cloud service providers. This directly complies with the mandatory regulations of many countries that require sensitive geographical and personal information to be stored locally.
After the data is stored in overseas local data centers, the next key steps are "data preprocessing" and "compliant cross-border transfer". Automakers will use the local teams and platforms of their partners to desensitize and annotate the data stored in their partners' cloud. By technical means, personally recognizable information such as facial images and license plates is erased, leaving only key features such as driving scenarios and behaviors for algorithm training.
When such data is needed for model training, instead of transmitting terabytes of original video or radar point cloud data, the "feature data" or "training sets" that have undergone desensitization, annotation, and preliminary model preprocessing are transmitted across borders. This processed data package significantly reduces transmission costs and time, and its content complies with the cross-border flow requirements of data protection regulations such as GDPR.
Case of XPeng P7 European Version
XPeng has used the regional data centers of Amazon Web Services (AWS) in Europe to build a complete set of independent back-end services for its Internet of Vehicles (IoV) platform targeting the European market. For all XPeng vehicles sold in Europe, the data is transmitted and stored directly on AWS servers located within the EU through secure Internet of Things (IoT) channels from the moment of collection.
XPeng uses AWS IoT services to manage the connections of tens of thousands of vehicles, Amazon S3 to store massive driving data, and data processing services such as EMR and EKS (Elastic Kubernetes Service) to process, analyze, and train models on these data locally in Europe.
XPeng has established a "European data security domain" that is isolated from its business in Chinese Mainland both physically and logically. The data of European users - from generation, transmission and storage to processing and final destruction - is totally completed within this closed loop, thus ensuring compliance with GDPR at the architectural level.
Chinese intelligent driving suppliers license their core technologies such as software algorithms, which have been verified with massive data, to overseas automakers or mobility platforms. The two parties jointly conduct secondary development and adaptation for specific scenarios in the target market to ensure the localization and compliance of the technical solutions overseas. Chinese intelligent driving suppliers can also receive feedback of overseas road data.
Most intelligent driving companies going overseas have adopted a "two-legged" strategy, that is, promoting the R&D and implementation of both L2/L2+ and L4 simultaneously. The former can quickly generate cash flow and accumulate experience through mass production cooperation with OEMs; the latter cooperates with mobility platforms to operate Robotaxis in specific areas (Operational Design Domain, ODD) to obtain massive driving data at low cost.
In the future, in addition to technology licensing, there may be more in-depth binding models such as the establishment of joint ventures and equity investment to jointly explore overseas markets.
Cooperation Case between Momenta and Uber:
Uber's global platform is used as a commercial outlet for Momenta's intelligent driving technology. The two parties have jointly launched Robotaxi services in markets other than China and the United States, and plan to officially put them into operation next year. Europe, especially Munich, Germany, is the first stop and test site of this overseas layout plan. Their cooperative test fleet has already taken to road.
Uber has 150 million monthly active users in Europe, which greatly reduces the difficulty and cost for Momenta to explore the overseas market independently. Moreover, the actual operations in Europe not only allow to obtain valuable overseas road data to feed back into its technical algorithms, and accelerate the operation of the "data flywheel", laying the foundation for the future overseas layout of the "other leg" (referring to the R&D and implementation of L4), but also help to enhance its international value and evaluation in the capital market.