Breaking Barriers with China Speed in Intelligent Driving: - Gao Feng Advisory

CN
MENU

Breaking Barriers with China Speed in Intelligent Driving:

release time：2025-03-01

Breaking Barriers with China Speed in Intelligent Driving

GAO FENG ADVISORY
March 2025

I. Current Situation and Challenges: Homogenization Trend and Commercial Closed-Loop Dilemma

Intelligent driving technology is evolving from a "feature-stacking race" to a "cost optimization" phase, yet the industry faces three core challenges:
  1. Convergent technical paths: Most models adopt the BEV + Transformer architecture and similar cockpit interaction designs;
  2. Homogeneous functional configuration: L2+ penetration has exceeded 80%, with features like automatic parking (APA) becoming standard;
  3. Commercialization bottleneck: Despite BYD driving down system costs, automakers struggle to monetize intelligent driving capabilities, trapped in the dilemma of "no intelligent driving, no sales; with intelligent driving, no profit".
BYD’s "technology equality" initiative has accelerated market adoption of intelligent driving but eroded its functional premium. Technological progress has not yet translated into a sustainable business model that balances consumer demand and profitability for automakers.

II. The Future of Driving: From Single-Vehicle Intelligence to a Human-Centered Intelligent Travel Ecosystem

AI is widely applied in autonomous driving data training, improving annotation and model training efficiency, and replacing manual data collection/road mapping. However, deep learning relies heavily on high-quality labeled data; the complexity and variability of driving scenarios lead to high data collection costs and limited model generalization. Safety concerns and high computational requirements restrict large-scale deployment, while high-performance chips drive up costs and power consumption, raising mass production barriers.

The Value of DeepSeek and Model Optimization

The emergence of DeepSeek has partially mitigated these shortcomings: its model supports distributed training and edge computing deployment, enabling low-latency reasoning even on low-computing-power chips, and providing cost-effective solutions for large-scale intelligent driving. With the optimization of distillation technology, vehicle model volume and energy consumption are expected to drop further—migrating large model knowledge to lightweight models reduces hardware load while retaining core capabilities, making it more suitable for in-vehicle environments and advancing high-level autonomous driving.

Limitations of Single-Vehicle Intelligence and the Need for ITS Coordination

The two mainstream single-vehicle intelligence paths—end-to-end models + neural networks (strong AI) and modular models + expert rules (weak AI)—both have performance ceilings. Single-vehicle intelligence can enhance individual vehicle perception and decision-making but cannot fully solve urban traffic’s overall efficiency and safety issues. Thus, intelligent driving development must move beyond single-vehicle intelligence and evolve in coordination with Intelligent Transportation Systems (ITS) for large-scale breakthroughs.

Evolution to City-Level Systematic Intelligent Transportation

Genuine traffic intelligence relies on deep interconnection of people, vehicles, roads, and clouds to improve the transportation system’s overall safety and efficiency. AI-powered intelligent transportation covers traffic management, autonomous driving, vehicle-road collaboration, and other key areas, driving systemic upgrades:
  • AI optimizes traffic flow, alleviates congestion, and improves road safety via real-time data analysis and prediction;
  • Vehicle-road cooperative technology connects vehicles, urban infrastructure, and other traffic participants, providing accurate and efficient solutions for ITS.
In the future, the deep integration of single-vehicle intelligence and intelligent transportation will transform the system from vehicle-centered intelligence to city-level systematic intelligent transportation. This shift will reshape urban travel, making it safer, more efficient, and sustainable, and ultimately contribute to smart city development.

III. Bottlenecks and Possible Breakthroughs in Intelligent Transportation Development

Current development should focus on optimizing existing scenarios with mature technologies, with industry explorations including: smart bus stations + smart bus customer service; flexible bus + TOD mode; traffic police-intelligent mobility collaborative traffic control.
Despite these efforts, full-scale intelligent transportation faces multiple bottlenecks, with clear exploration directions for breakthroughs:

1. Lack of a Clear Legal and Regulatory Framework

A comprehensive legal/regulatory framework is critical for advanced autonomous driving commercialization, requiring the government to establish unified access standards, road permit procedures, and accident liability rules—unified regulations mitigate fragmentation and provide a stable policy environment. Without legal protection, mature technology cannot achieve commercial operation (e.g., Singapore’s 2017 Road Traffic (Amendment) Act clarified autonomous driving test legality, responsibility, and road access, providing efficient policy guarantees).
China’s current status: National and local policies/standards have formed an initial regulatory framework for road testing, demonstration applications, and supervision, but with significant regional implementation differences and complex approval procedures. The 2021 Management Specifications for Road Testing and Demonstration Application of Intelligent Connected Vehicles (Trial) standardized local rules but did not fully clarify autonomous driving-specific regulations and safety standards, leading to legal obstacles for enterprise pilots. Local government support and implementation details also vary, resulting in uneven pilot progress.
Breakthrough direction: The current MIIT-led model has implementation blind spots; future exploration can focus on the Ministry of Public Security taking the lead in formulating more practical top-level design and licensing implementation plans.

2. Absence of a Data Collection and Sharing Mechanism

Intelligent transportation systems rely on high-quality real-time data, but a comprehensive underlying data collection/sharing mechanism has not been established, leading to fragmented information collection, difficulty in obtaining complete travel chains and activity spatio-temporal rules, and directly limiting AI/digital technology potential.
Key issues:
  • Traffic data collection involves multiple departments/systems, but independent collection (lack of standardized protocols/regulations) leads to inconsistent data formats and ineffective integration/sharing;
  • Road traffic accident information collection mainly serves handling/liability identification, lacking correlation analysis of driver/vehicle/road/environment data, failing to meet big data-era statistical analysis needs;
  • The lack of top-level design makes urban traffic big data platforms unable to form cognitive insight, deductive judgment, and large-scale calculation capabilities for complex traffic laws, restricting planning and real-time regulation;
  • Severe information silos waste data resources and limit overall system efficiency.
Breakthrough direction: Build a machine learning-adaptable system with core data and key elements, iterate and improve via a "small steps, fast running" strategy; continue to promote basic data collection and sharing, providing standardized, high-quality data support for vehicle-side and other terminal devices.

3. Lack of Technical Standardization and Interoperability

Intelligent transportation systems lack unified technical standards across regions and manufacturers, resulting in poor system interoperability and severely restricting overall synergy, especially data interoperability.
Key issues:
  • Diverse communication protocols: DSRC and C-V2X (the two major V2X technologies) have not achieved full interoperability;
  • Inconsistent data formats/interface specifications: Increase the complexity of data sharing and system integration;
  • Spectrum allocation discrepancies: Cross-border/cross-regional system communication and interoperability challenges due to national differences;
  • Differing safety standards: Lead to varying system security/reliability and higher interoperability difficulty.
Breakthrough direction: Formulate unified technical standards for device interoperability via joint efforts of the government, industry organizations, and enterprises, promoting coordinated intelligent transportation development and improving overall traffic efficiency and safety.

4. Urgent Need for Data Security and Privacy Protection

Data security and privacy protection are key challenges for data sharing in intelligent transportation systems. A weak data security mechanism may cause leakage/improper use, eroding public trust; new technologies and increased connectivity also create malicious attack vectors and privacy risks (e.g., theft/tampering of vehicle location, driving route, and other sensitive data during transmission/storage).
Breakthrough direction: Establish a sound data security mechanism to ensure data safety during collection, transmission, and sharing, preventing leakage/improper use. This requires both technical measures (encryption, anonymization) and sound laws/regulations/supervision to enhance public trust in intelligent transportation systems.

IV. Scenario Analysis and Key Players in the Future Development of Intelligent Transportation

At the current stage, clear policy frameworks, high-quality data-driven AI potential, and technological breakthroughs are the core drivers of intelligent transportation development. Policies provide legal guarantees and R&D/operation direction for enterprises, while high-quality data collection is the foundation for effective AI application and future technological progress.
Driven by the two core factors of policy and technology, the intelligent transportation industry will evolve along four scenarios, with distinct key players and focus areas for each:

Scenario 1: Current Situation – Partial Policy Support + Immature Technology

  • Policy: The government has introduced some supportive policies, but delays, insufficient coverage, and incomplete national policies (technology development, market access, regulatory standards) create industry uncertainty; local pilots exist but progress is uneven.
  • Technology: AI-powered autonomous driving and V2X have made minor breakthroughs but are far from mature, especially in handling complex traffic environments.
  • Key players: OEMs and intelligent driving software/hardware suppliers (the primary driving forces). OEMs cooperate with local governments to promote autonomous driving pilot applications, test technology feasibility, and drive market development; mobility service providers are limited by technological bottlenecks and policy delays, with traditional ride-hailing models still dominant.
  • Industry trend: Market penetration of intelligent connected vehicles gradually increases as policies/technologies improve, but large-scale autonomous driving commercialization is difficult to achieve.

Scenario 2: Efficient Interconnected Mobility – Partial Policy Support + Mature Technology

  • Policy: The government has completed top-level design, but regional differences, road complexity, and policy implementation delays prevent effective enforcement.
  • Technology: AI-powered autonomous driving has made significant breakthroughs, greatly improving intelligent driving reliability/safety and enabling accurate, stable performance in complex road conditions.
  • Key players: OEMs and their suppliers (still the core driving forces). Automakers demonstrate AI intelligent driving’s safety/reliability to advocate for policy relaxation and accelerate implementation; consumer acceptance of intelligent vehicles gradually increases, providing momentum for policy breakthroughs.
  • Industry trend: Private car ownership remains dominant; intelligent vehicles achieve seamless connection across travel scenarios (e.g., autonomous parking) and share real-time data with urban infrastructure (smart street lights, traffic signals) to optimize traffic flow, delivering a more efficient, personalized travel experience.

Scenario 3: Progressive Intelligent Mobility under Policy Guidance – Fully Supported Policy + Immature Technology

  • Policy: The government has introduced proactive favorable policies, improved regulations, and created a sound industry environment, playing a key role in promotion.
  • Technology: AI-powered autonomous driving still faces significant bottlenecks; high-level intelligent driving requires more time and technological progress.
  • Key players: OEMs, technology suppliers, and mobility service providers (cross-industry collaborative innovation drivers).
  • Industry trend:
    • OEMs increase autonomous driving R&D investment under policy support to drive technological breakthroughs;
    • Mobility service providers accelerate the promotion of autonomous driving MaaS mode, developing shared transport platforms and gradually changing traditional travel modes;
    • Private cars and autonomous driving shared transport coexist (especially in high-demand large cities); consumer acceptance of intelligent transportation increases, and the boundary between private and shared transport blurs.

Scenario 4: Human-Centered Intelligent Mobility Ecosystem – Fully Supported Policy + Mature Technology

  • Policy: The government has established a comprehensive autonomous driving regulatory framework, proactively promoting intelligent transportation infrastructure construction and the large-scale application of V2I/V2X technologies.
  • Technology: AI-powered autonomous driving breakthroughs have greatly improved intelligent transportation system reliability/safety; intelligent driving and related technologies have become daily applications.
  • Key players: Mobility service providers (the core of the ecosystem); OEMs shift their role from C-end direct sales to B-end customized intelligent vehicle supply.
  • Industry trend:
    • The MaaS model becomes mainstream, enabling users to select/pay for multiple transportation services via a single platform for seamless connection and improved travel efficiency;
    • Intelligent shared transport becomes more attractive, and some consumers abandon private cars for shared services;
    • In-depth OEM-mobility service provider cooperation becomes an industrial development key; autonomous driving fleet intelligent operation emerges as a new market growth point.

Long-Term Industrial Evolution and Smart City Integration

Intelligent transportation will become a core element of urban development, improving social resource efficiency and reshaping human-city and human-environment interactions. With the development of the low-altitude economy, intelligent transportation will break ground limitations, and the coordination of roads, airspace, and rail will create a multi-layered transport network, revolutionizing urban commuting:
  • Future urban planning will shift from road widening/rail expansion to intelligent dispatching system development, realizing dynamic optimization of ground transportation, low-altitude drone delivery, air taxis, and other multi-dimensional modes;
  • Urban transportation systems will gradually evolve from "rule-driven" to "data-driven" with the improvement of AI decision-making capabilities, achieving more efficient transportation resource allocation and management, and fundamentally changing urban planning logic to build more flexible, sustainable smart cities.

Core Conclusions and Industry Implications

  1. Policy and technology are the core determinants: Policy implementation and subtle changes shape the industry’s future direction, while continuous technological evolution (especially AI breakthroughs such as DeepSeek) provides new growth momentum, reshaping the competitive landscape and development path;
  2. Forward-looking insight and agile strategy are critical: In a rapidly changing market/technological environment, decision-makers must accurately identify policy/technology trends and adjust strategies quickly in uncertainty;
  3. Scenario-based investment and industrial layout: Future investment will focus on the precise matching of different development scenarios, promoting the deep integration of intelligent driving and intelligent transportation;
  4. Beyond industrial transformation: Technological progress will drive social development model changes, reshaping travel methods, production systems, urban operation mechanisms, and the relationships between people and technology, and cities and transportation.

About the Authors

  1. Edward Tse: Founder & CEO of Gao Feng Advisory, a pioneer in China’s management consulting industry. He led the Greater China operations of BCG and Booz for 20 years, providing consulting services to hundreds of Chinese/foreign companies, investors, start-ups, public-sector organizations, Chinese government agencies, the World Bank, and the Asian Development Bank. Author of hundreds of articles and seven books (e.g., The China Strategy (2010), China’s Disruptors (2015), Mindset for Mega Changes 2 (2025)).
  2. Lu Zhang: Senior Manager at Gao Feng Advisory (Shanghai), with deep expertise in strategic and financial consulting. Specializes in the automotive, mobility, and aviation sectors, focusing on strategic planning, financial modeling, and investment analysis. Excels at translating strategy into financial execution, providing forward-thinking, results-driven solutions for enterprise growth and transformation.



Gao Feng Advisory 

Gao Feng Advisory Company  is a professional strategy and management consulting as well as investment advisory firm with roots in China coupled with global vision, capabilities, and a broad resources network

Wechat Official Account：Gaofengadv

Shanghai Office

Tel: +86 021-63339611

Fax: +86 021-63267808


Hong Kong Office

Tel: +852 39598856

Fax: +852 25883499


Beijing Office

Tel: +86 010-84418422 
Fax: +86 010-84418423


E-Mail: info@gaofengadv.com

Website: www.gaofengadv.com

Weibo: 高风咨询公司

Previous： Businesses need to embrace an ‘AI mindset’

Next： China’s social-minded technopreneurs are a force to reckon with