The father of autonomous driving: Humanoid robots are both overestimated and underestimated, and the market potential for aerial robots will far exceed that of ground robots

At Morgan Stanley's 24th Asia-Pacific Summit, renowned analyst Adam Jonas had an in-depth conversation with Sebastian Thrun, known as the "father of autonomous driving." As the founder of Google X and the creator of the predecessor project to Waymo, Thrun's insights helped investors clear the fog surrounding the fields of AI and robotics.

According to the Wind Trading Desk, on November 20th, Morgan Stanley released the Minutes. The core conclusions of this dialogue are very clear, directly addressing two major pain points in the current market:

1. Potential Revaluation of Tesla: Thrun clearly pointed out that if Musk can achieve safe Robotaxi commercialization in Austin through a "pure vision" (relying solely on cameras) solution, it would be an extremely significant breakthrough. This essentially implies that once the pure vision solution is operational, the costly lidar route will face severe challenges in cost efficiency.
2. Structural Differentiation in the Robotics Field: Humanoid robots are in a contradictory state—severely overestimated due to their enormous potential market size (TAM), but at the same time, the actual difficulty of solving open-ended tasks like hand dexterity is severely underestimated. Moreover, crucially, Thrun pointed out that the future growth of robots will be primarily "in the air" rather than on the ground, with the number of aerial robots far exceeding that of ground robots, making upgrades to air traffic control systems urgent.
3. Stage of the Industry: Although investors feel anxious, science has only solved 1% of the problems, with 99% still to be explored. Autonomous driving is on the verge of transitioning from L4 to L5.

For investors, this means that in the short term, they need to closely monitor Tesla's pure vision FSD performance in Austin, which is not only a technical test but also a validation point for the business model. At the same time, in the hardware sector, they should be wary of the excessive hype around humanoid robot concepts and instead focus on underlying technology companies that can solve specific physical interaction challenges (such as dexterous hands).

The "Wright Brothers Moment" of Autonomous Driving Has Passed, and the Penetration Rate is at a Turning Point

Sebastian Thrun recalled that the "Wright Brothers Moment" for autonomous vehicles occurred back in 2005 when he won the DARPA Grand Challenge. Today, this technology is rapidly penetrating the market.

Data Verification: Among the approximately 500 attendees at the summit, a show of hands indicated that about one-third had already ridden in an autonomous vehicle, the vast majority of which experienced Waymo.

Market Potential: Humans spend a total of 82 million years in cars each year, which means the economic value of freeing up driving time remains enormous. Morgan Stanley predicts that the penetration rate of autonomous driving is on the verge of a full transition to L4/L5 levels.

Thrun provided a weighty judgment on the technological route. He specifically mentioned that if Elon Musk could demonstrate a passive optical (camera-only) Robotaxi in Austin and achieve safe commercial service without a driver, it would be a "huge" achievement This viewpoint implies to investors that the current market debate over multi-sensor fusion (LiDAR + radar + cameras) and pure vision solutions is far from over. Once pure vision solutions are proven to be safe and feasible, their cost advantages will pose a significant threat to competitors.

The Paradox of Humanoid Robots: Overestimated Market vs. Underestimated Technical Difficulty

Regarding the currently hot humanoid robot sector, Thrun presents a seemingly contradictory yet insightful perspective: They are both "overhyped" and "underappreciated."

Overestimated: The market has overly emotional expectations regarding the total potential market size for humanoid robots to replace human labor.

Underestimated: The market severely underestimates the actual engineering challenges of enabling robots to perform open-ended tasks and possess "dexterous hands." This indicates that there is still a significant technological gap to bridge before truly universal humanoid robots can be realized, especially in terms of physical interaction.

Aerial Robots Will Far Exceed Ground Robots, Infrastructure Needs Upgrading

The robots of the future will not be on the ground but in the sky. Thrun predicts that the number of aerial robots will greatly exceed that of ground robots.

Current Technology: The technology required for fully automated operations in 3D space already exists.

Infrastructure Bottleneck: The key limitation to the explosion of this field lies in infrastructure; the United States' air traffic control system (ATC) urgently needs significant upgrades to accommodate this trend. This provides a long-term investment rationale for sectors focused on eVTOL (electric vertical takeoff and landing aircraft) and related infrastructure development