Uber CEO on the Future of Autonomous Driving and the Ride-Hailing Business
How will autonomous driving transform Uber's strength in decentralized distribution—the ability to "aggregate demand and pour it into any mobility asset"? This article, based on the latest remarks by CEO Dara Khosrowshahi (summarized in the text), organizes the discussion into four axes: (1) differences in technical approaches, (2) partnership strategies and revenue models, (3) robotics/drone expansion in delivery, and (4) impact on employment.
1. Divergence in Technical Approaches: Camera-only vs. Sensor Redundancy + HD Maps
1-1. Tesla Approach vs. Waymo Approach
Dara contrasted the design philosophies of autonomous driving: "increasing software complexity to keep hardware light" (Tesla) versus "reducing the burden on software through sensor redundancy like LiDAR and radar plus HD maps" (Waymo and others). In fact, Tesla has shifted to a vision-centric (camera-only) approach, publicly stating its goal of cost reduction and scale.
1-2. Cost Curve: From "High-End Luxury" to Commodity for LiDAR
Long-range LiDAR costs have dropped from approximately $75,000 in 2015 to around $500 today. Solid-state LiDAR is expected to reach the $300 level. Scaling down hardware will lower unit costs on the network side (cost per ride/delivery) and accelerate the path to profitability.
1-3. Safety: "Surpassing Humans" is Becoming Realistic
Waymo has presented public data showing a 91% reduction in severe accidents and an 80% reduction in injury accidents compared to human drivers in cities where it operates, such as Atlanta and Austin. Peer-reviewed studies have also reported a significant decrease in injury accident rates. Dara's view that "autonomous driving will become 'superhuman' in the long term and save millions of lives" is supported by data, at least in current city-limited operations.
2. The Current State of "Platform x Partnership": Waymo Collaboration and the Rise of Chinese Players
2-1. Implementation with Waymo: Atlanta/Austin
Full-scale service where users can be matched with Waymo's fully autonomous vehicles via the Uber app began in Austin in March 2025 and in Atlanta in June 2025. Geographic coverage and fleet size are expanding sequentially. "Customers highly value the new, quiet, and 'slightly exciting' experience and return to use it again," says Dara.
2-2. International Expansion of Chinese AVs and Uber
Uber has signed a multi-year strategic partnership with Baidu (Apollo Go). The goal is to deploy thousands of ride-hailing vehicles in multiple markets outside of the U.S. and mainland China. Chinese players (Baidu, Pony.ai, WeRide, etc.) have honed their performance in complex urban traffic environments, and their global expansion will work in mutual complement with Uber's demand-side platform.
3. Changes in Revenue Models: Moving Toward "Vehicle Financialization" While Remaining Asset-Light
3-1. Redefining Network Effects
"Compared to an independent fleet, being on the Uber network results in shorter average pickup distances and higher utilization (paid mileage ratio)." The advantage of demand aggregation remains unchanged in the AV era. The "hybrid use" of direct (proprietary app) and marketplace models is an optimal solution already proven in food delivery.
3-2. "Financialization of Fleets"
Dara's view that "in ten years, REIT-like financial owners will hold AV fleets and put them on the network to maximize utilization" is consistent with a capital policy where Uber proves the model on its own balance sheet in the short term and moves it off-balance sheet after maturity. While generating over $8.5 billion in cash flow annually and pursuing a 2 trillion yen share buyback (announced in 2025), Uber is securing "offensive capacity" for AV and fleet investments.
4. Automation of the Last Mile: Differentiating Between Sidewalk Robots and Drones
4-1. Sidewalk Robots: The "Cute" Delivery Person Within One Mile
Serve Robotics is expanding commercially with Uber in areas like Los Angeles. With plans to deploy 2,000 units, they aim to optimize unit costs for short-distance, high-frequency use cases. Implementations with Cartken and Avride are also expanding.
4-2. Drones: Targeting suburban and low-density areas
Uber has formed a strategic partnership with Flytrex to begin drone delivery in U.S. pilot markets later this year. This is also the company's first drone investment, aiming to expand the delivery TAM along the "Z-axis" alongside sidewalk robots. "Robot-to-drone" collaboration between Serve and Wing (Alphabet) is also progressing.
4-3. "Aerial Ride-Sharing": Laying the groundwork for eVTOL
Joby Aviation plans to integrate the air mobility services of the acquired Blade into the Uber app. The demonstration phase for capturing the "third dimension" of cities is approaching.
5. The Future of Employment: Coexistence in the short term, social challenges in the long term
“We can absorb this through natural turnover over the next 5–7 years. Growth is fast, and robot deployment can be offset by tightening hiring. It will be a full-scale challenge in 10–15 years,” says Dara. Uber is exploring new on-demand jobs like AI labeling and the value of human labor in the "first/last few dozen meters" of delivery. It will also continue to cooperate with cities and regulators (supporting the design of charging networks and pick-up/drop-off points).
6. Summary: The "Winning Strategy" in the AV Era
Product side: Proven safety records in limited urban areas (Waymo) and declining LiDAR costs are the foundation for accelerating adoption.
Go-to-market: Superimposing AV fleets onto Uber's demand-side network to optimize utilization rates and unit prices. Scaling through over 20 partnerships with companies like Waymo and Baidu.
Last mile: Expanding the delivery TAM by dividing roles between sidewalk robots (~1 mile) and drones (suburbs).
Capital policy: High cash flow plus large buybacks combined with selective AV and fleet investment. The endgame is a financial owner-style "fleet financialization."
“We are the 'heart of the network' that pours demand into assets. Once the conditions for safety and unit price are met, the winner in AV will be 'network x multiple fleets' rather than a 'single app.'”
— Dara Khosrowshahi (Summary of remarks, from the main text)
