As the halo fades and rationality returns, autonomous driving has entered a new stage of "pragmatism".
At the beginning of 2026, the global autonomous driving sector continued to advance in technological validation and commercialization exploration. The approval of the first batch of Level 3 conditionally autonomous vehicles in China, Tesla's plan to start mass - producing vehicles without steering wheels and pedals in the second quarter, and Waymo's continuous expansion of its driverless taxi service network mark that the focus of competition in autonomous driving is shifting from overcoming the technical challenges of "whether it can be achieved" to how to build a profitable, safe, and widely accepted ecosystem.
Recently, the McKinsey Center for Future Mobility released the third edition of its Autonomous Driving Industry Leaders Survey. This report, based on insights from global industry experts, reveals a key trend: after the early enthusiasm, the autonomous driving industry is shedding its glamour and entering a new development stage that is more pragmatic, rational, and full of complex challenges.
Timelines are generally postponed: Commercialization expectations become more conservative
The most obvious finding in the report is that the industry's expectations for the popularization speed of autonomous driving have significantly declined. Compared with the 2023 forecast, the 2025 survey report postponed the popularization timelines of most autonomous driving application scenarios by an average of 1 - 2 years. Although some cities in China and the United States have launched Level 4 driverless taxi services, the expected time for large - scale global commercial promotion has been postponed from 2029 to 2030. The Level 4 urban pilot for private passenger cars is expected to be postponed from 2030 to 2032, and the commercial viability of fully autonomous trucks has also been postponed from 2031 to 2032. Overall, experts believe that Level 4 autonomous driving will first be commercially applied in the taxi sector.
Regional differentiation intensifies: China and the United States lead, and echelons emerge
The development process of autonomous driving also shows regional differences. China and the United States have become pioneers in commercialization due to their faster development cycles, active capital and startup ecosystems, proactive regulatory attempts, and more suitable road environments. Deployments in other regions are relatively lagging. The surveyed experts expect that it will take 3 to 7 years for autonomous taxis to be widely commercially deployed in all regions of the world. The truck and passenger car sectors are expected to show a similar regional promotion model, with China and the United States expected to significantly lead other regions in Europe or Asia in most application scenarios.
Market focus shifts back: From a "leap" in Level 3 to a "gradual" approach in Level 2+
The industry's focus on the private passenger car market has shifted back from Level 3, which "liberates hands", to Level 2+ (enhanced advanced driver - assistance systems) dominated by humans.
49% of the surveyed experts believe that by 2035, the core of the mass market for private passenger cars will focus on Level 2+ functions. Previously (in the 2023 survey), 52% of experts expected the market to develop towards Level 3 and more advanced systems. This change in expectations is due to real - world challenges such as the slower - than - expected decline in the cost of Level 3 systems, high development and verification costs, and slow technological progress. Currently, only 39% of experts still believe that Level 3 will become the market center. Experts generally believe that Level 3 and above functions are being re - positioned as optional or featured functions for high - end models.
Cost expectations are raised: The commercialization threshold for high - level autonomous driving increases
Contrary to the common perception that technological progress is accompanied by cost reduction, experts have significantly raised their cost estimates for achieving Level 4 and above autonomous driving. This is especially evident in areas such as autonomous trucks that have not yet been scaled up, where cost expectations have been raised by 50% - 60%. The reasons include huge investments required for industrial production and verification of a large number of "long - tail" edge scenarios when transitioning from testing to commercialization.
In contrast, the cost estimates for Level 4 driverless taxis are more stable due to their scale deployment. Analysis shows that the software R & D and verification costs for low - level autonomous driving can be 4 to 7 times lower than those for high - level systems, and the software investment required for full - scale driverless operation may exceed $3 billion.
Core pain points are clear: Cost pressure surpasses technology as the primary challenge
The ranking of pain points in the current stage of industry development has changed fundamentally. The survey shows that high costs have become the most prominent pain point in the ADAS development process, surpassing technology itself and liability attribution issues. This confirms that the industry is moving from being "R & D - driven" to being "deployment and cost - control - driven". Product liability and regulatory uncertainty are ranked as medium - level pain points, reflecting the urgent need to establish a clear industrial liability framework. Market competition and talent scarcity are ranked lower among the pain points, suggesting that the value pool after industry consolidation is large enough to accommodate multiple participants, and AI tools have alleviated the talent bottleneck to some extent.
Regionalization of technology stacks: An independent Chinese ecosystem becomes the mainstream expectation
When asked about the possible development paths of the global ADAS technology stack, the surveyed experts had different views. 74% of experts predict that China will form an independent technology stack, which is due to Chinese consumers' high attention to ADAS functions, a complete local supply chain system, and the geopolitical background. Regarding the global pattern, experts' views are divided into three categories: China and the United States will each have an independent technology stack (26%); China will have an independent technology stack, while the United States and Europe will share another technology stack (35%); the United States will have its own technology stack, and another technology stack will be developed by China and shared with Europe (12%).
Consensus on technology path: Hybrid architecture becomes a pragmatic choice
Regarding the cutting - edge "end - to - end" learning driving system, experts affirm its potential performance and cost - effectiveness, which can bring a more human - like driving experience and lower R & D costs - about two - thirds of experts predict that development costs can be reduced by more than 10%. However, they have serious concerns about the safety and regulatory challenges caused by its "black - box" characteristics. Therefore, only 22% of experts believe that the "pure end - to - end" model will become the mainstream. The vast majority (78%) of experts believe that the future belongs to the "hybrid architecture" - integrating "end - to - end" AI models with traditional interpretable and verifiable rule - based algorithms to ensure safety redundancy while pursuing performance.
Strategic suggestions: Build agility and resilience in the face of uncertainty
Facing the new stage, the report puts forward key suggestions for industry participants:
Maintain extreme agility: Have the ability to quickly perceive and adjust in the face of rapid changes in technology, regulations, and costs.
Focus on the core and cooperate openly: During the industry consolidation period, deeply cultivate the most competitive value - chain segments and build ecological advantages through cooperation.
Adhere to customer - value orientation: Return from being technology - driven to solving real user pain points and pay attention to users' actual willingness to pay for functions.
Collaborate to build safety and standards: Actively cooperate with regulatory agencies to promote the establishment of clear safety standards and liability frameworks to pave the way for large - scale development.
Strengthen organizational collaboration: Break down the barriers between technology, business, and operations to cope with cost and long - tail challenges with system efficiency.
Judging from the repeatedly postponed timelines, rising costs, and the pragmatic choice of technology routes, the road to large - scale popularization of autonomous driving is longer and more complex than expected. However, there is a clear consensus in the industry: in the private passenger car sector, the next decade will be dominated by the gradual improvement of Level 2+; in the commercial sector, driverless taxis and autonomous trucks will focus on specific regions and scenarios and gradually expand their business territory.
Source: McKinsey
This article is from the WeChat public account "Automotive Market Insights". The author is Zheng Li. It is published by 36Kr with permission.