Autonomous Driving Accelerates into Reality
Autonomous vehicles are entering reality at an unprecedented pace. Over the past few years, the first L3 vehicles hit the market, weekly fully autonomous taxi trips exceeded 700,000, and autonomous trucks also completed their first driverless demonstrations. This progress is global: Europe has launched more than 35 autonomous driving pilot projects, while the United States and China provide more than 450,000 and 250,000 commercial trips per week, respectively.
The industry's rapid advance does not mean the competition is over. Developers are still working to overcome the next round of technology and investment barriers, and there is no consensus yet on how the technology stack will evolve. In other words, advanced driver assistance systems and autonomous driving still have significant room for scaling.
Adoption Timelines Generally Delayed
The McKinsey Center for Future Mobility, together with The Autonomous, conducted a biennial survey of 91 global industry decision-makers. The results show that, compared with the 2023 survey, the adoption timelines for most autonomous driving use cases have been delayed by one to two years on average. For example, large-scale global deployment of L4 robotaxis is not expected to become a reality until 2030, compared with the previous expectation of 2029; L4 private passenger vehicle urban pilots are expected to be delayed from 2030 to 2032; and fully autonomous trucks are not expected to reach commercial viability until 2032, rather than the previously expected 2031.
The survey also shows significant differences in deployment pace across regions. The United States and China have already taken the lead in launching robotaxis, but other regions may need three to seven years to achieve widespread commercial deployment. Experts expect trucks and passenger vehicles to follow a similar pattern: China and the United States lead, with most use cases launching significantly earlier than in Europe and other parts of Asia. This regional difference is influenced by multiple factors, including faster development cycles, flexible business models and startup culture, regulatory support, availability of capital, strong AI and software foundations, a more autonomous-driving-friendly built environment, larger market size, and a stronger willingness to test new technologies at scale.
Mass Market for Private Cars Shifts to L2+ Features
Among experts surveyed this year, the largest group (49%) believe that by 2035, the mass market for private cars (not robotaxis) will focus on L2+ features. This contrasts sharply with the 2023 survey, when 52% of experts expected the market to shift to L3 or higher-level systems. Previous forecasts were mainly based on assumptions that L3 system costs would decline, as well as the advantages of L3 vehicles allowing drivers to rest or engage in non-driving tasks such as work and gaming.
However, only 39% of experts now predict that the market will focus on L3 or higher. This decline may stem from a lack of funding, high development and verification costs, and slow technological progress. Experts increasingly view L3 autonomous driving as an optional niche product, mainly for high-end models, rather than a broad mass-market offering.
Cost Estimates for Higher-Level Autonomous Driving Rise## Cost Estimates for Higher-Level Autonomous Driving Rise
Compared with 2023, experts' estimated costs for higher-level autonomous driving have risen, despite additional breakthroughs in the AI field. This may be because the costs of transitioning from testing to commercial deployment have been underestimated—such as product industrialization, the ability to handle long-tail edge cases, and high verification and validation costs. The survey shows that cost estimates for Level 4 robotaxis have increased only slightly, possibly because robotaxis are already in the scaling stage, and leaders are more certain about what is needed to achieve commercial maturity. In contrast, cost estimates for autonomous trucks have been revised upward more significantly—by 50% to 60% depending on the use case.
Lower levels of autonomous driving typically require relatively simple algorithms and streamlined hardware configurations (sensors and computing). Experts believe this could make software development, testing, and validation costs four to seven times lower than for higher levels (Level 4 and Level 5 robotaxis and trucks). On the other hand, urban robotaxis and other highly autonomous vehicles must cope with complex traffic scenarios and a wide variety of edge cases. These technical hurdles mean that robotaxis and "full-journey" autonomous trucks require software investments of more than $3 billion to reach market readiness. Hardware development costs appear to be related to vehicle form rather than level of autonomy, with trucks (whether used for hub-to-hub or full-journey operations) having significantly higher hardware costs than other vehicle types.For decision-makers and practitioners, the key is no longer pursuing the most cutting-edge autonomous driving level, but finding a balance among cost, technology, regulation, and market acceptance. The latest insights from experts provide important pointers for understanding this complex and rapidly evolving industry.