Uber vs. Tesla, Robotaxi Timelines, and the End of Human Driving | Uber CEO Dara Khosrowshahi | #243

Summary

• Uber’s robotaxi strategy is to aggregate demand and orchestrate the ecosystem, not insist on one autonomous stack. Dara Khosrowshahi expects hybrid fleets of human and robot drivers, with Waymo as a leading partner and more than 20 autonomy partners; Uber targets 15 cities by year-end and, by 2029, more facilitated autonomous and robotaxi rides than anyone else.
• The operating moat is everything surrounding the autonomous driver. Uber supplies demand, data, pickup and drop-off intelligence, fleet management, maintenance and cleaning, letting partners focus on a safe, affordable driver; longer term, Khosrowshahi sees Blackstone-like owners owning fleets for perhaps a “9% yield” while Uber operates them.
• Tesla is optional to the thesis, not essential to it. Uber already has tens of thousands of Teslas on-platform, some using FSD, and would welcome autonomous Teslas once the camera-only approach is safe; whether the companies partner remains “TBD,” while Khosrowshahi expects “many, many winners.”
• Autonomy can become standard in new cars quickly while taking decades to dominate the installed fleet. Khosrowshahi expects every new car sold within 10 years to carry autonomous software and a sensor stack, but the average US car lasts more than 10 years and adoption across Uber’s 70-plus countries will vary; ultimately, falling trip costs should make private ownership “just not going to make sense.”
• Driver displacement is not Khosrowshahi’s base case through 2030. With Uber’s business growing over 20% and its audience almost 20%, he expects significantly more drivers, managing autonomy through natural annual churn of around 20% and slower recruitment; faster growth in Austin and Atlanta suggests robotaxis might expand demand rather than merely substitute labor.
• Uber’s larger TAM is “wiring up things that move,” while autonomy itself could be another trillion-dollar marketplace. That includes one-hour retail delivery, trains and boats, an Uber-to-Joby-to-Uber journey, suburban drones and urban sidewalk robots.
• The host-framed turnaround—from losing about $4.5 billion annually to earning more than $10 billion—rests on concentrated execution. Khosrowshahi’s maxim is that “the noes amplify the yeses,” paired with an unapologetically intense culture for people who “run to the fire” and want difficult, consequential work.

Deep dive

1. Focus turned Uber around—and still governs its moonshots

• Peter Diamandis framed the turnaround in stark terms: Khosrowshahi left a $200 million Expedia stock package to inherit an Uber losing roughly $4.5 billion annually; today, the host said, it earns more than $10 billion. Khosrowshahi’s diagnosis was organizational: Uber had been “guilty” of attempting too much.

• His strategy maxim, credited to Barney Harford: “Strategy is just as much about what you’re doing and what you choose not to do.” Saying yes is easy; the difficult “noes amplify the yeses,” concentrating the company behind one direction.

• The same discipline explains selling Uber Elevate to Joby when flying cars remained “a decade away,” then partnering with specialized hardware makers after Uber’s catalyst helped bring funding into the market. Internally, the bargain is explicit: flexibility exists, but Uber attracts people who “run to the fire”; impact has no “free lunch,” and employees must “work your ass off.”

2. The robotaxi race becomes a marketplace, not a winner-take-all duel

• Khosrowshahi rejected an all-autonomous-or-none outcome. Deployment will be hybrid: Waymo is a leader and already an Uber partner in Austin and Atlanta, while city fleets continue mixing autonomous vehicles with many human-driven ones.

• Uber wants “every terrific licensed robot driver” alongside every terrific licensed human driver. It has more than 20 autonomy partners—including Waymo, NVIDIA, WeRide, Pony.ai and Wabi—and expects to be in 15 cities with partners by year-end, aiming to facilitate more autonomous and robotaxi rides than anyone else in the world by 2029.

• Tesla remains welcome but unresolved. Uber has tens of thousands of Teslas on-platform and data from drivers using FSD; when camera-only Teslas are safe, “we’d love to have those Teslas,” Khosrowshahi said, but a partnership is “TBD.”

• Peter Diamandis’s platform question exposed Uber’s operating advantage: “Humans are much more complicated than robots.” Standard APIs can connect autonomous drivers, while Uber contributes training data, pickup and drop-off knowledge, demand aggregation, cleaning, repair and fleet management—services that human drivers previously handled themselves.

3. Robotaxi economics migrate from vehicle ownership to fleet yield

• Diamandis pressed the unresolved capital question with figures of roughly $150,000 for a Waymo vehicle versus Cybercab’s stated $30,000. Khosrowshahi did not address either figure directly; he forecast 10-plus vehicle providers on Los Angeles streets a decade out, while allowing that autonomous-software suppliers might consolidate.

• Uber will initially buy vehicles, but the intended destination is asset-light. Khosrowshahi’s analogy was Marriott: financial players own the hotels while Marriott manages them; similarly, “the Blackstones of the world” could own autonomous fleets producing perhaps a “9% yield,” while Uber operates and maintains them.

• Liability laws differ by jurisdiction. For autonomous vehicles, Khosrowshahi said liability would look more like Waymo being responsible for its own software driver, while Uber currently provides commercial insurance for human drivers on its platform. He is “quite confident” autonomous drivers will become much safer than humans, bringing industry liability costs down and creating savings that can reach consumers.

• Within 10 years, he expects every new car to have autonomous software and a sensor stack as lidar and camera costs fall. Fleet conversion will take much longer because the average US car lasts over 10 years and developing markets lag—but eventually lower trip costs and greater safety mean ownership is “just not going to make sense.”

4. Uber is wiring together every mode of movement

• Khosrowshahi described Uber as being “in the business of wiring up things that move”: people, food, groceries, retail, trains and boats. Invoking a VC friend’s “power of human laziness,” he said delivery beyond restaurants—from Best Buy to Sephora—has grown much faster than expected, with anything delivered within an hour “just exploding.” He also sees autonomy as another trillion-dollar marketplace and an opportunity to expand mobility and delivery TAM.

• The Middle East is an early launchpad in his view: WeRide vehicles are already operating with Uber in Abu Dhabi, while the UK offers promise through Wave’s OEM-agnostic software. Toward year-end, Uber hopes an Abu Dhabi customer can ride to a vertiport, fly Joby, then take another Uber; US service might follow next year, subject to licensing, production, experience and AI oversight.

• Delivery autonomy is harder because food does not get itself to and from the vehicle. Uber’s near-term answer is multimodal: drones could turn suburban orders into a 10-to-15-minute “surprise and delight,” while Coco and other sidewalk robots handle urban trips within a mile or two; specialized robots for bike lanes or roads may follow.

5. Autonomy changes the work mix before it eliminates workers

• Peter Diamandis framed Uber as potentially shifting value from labor to capital: drivers could become autonomous-vehicle owners or fleet managers. Khosrowshahi answered by defining Uber simultaneously as transportation infrastructure and “a platform for work,” already spanning driving, delivery, shopping and Uber AI Solutions assignments such as data labeling and model comparison.

• Khosrowshahi’s stated transition plan is to create work that leans into technology and give drivers an opportunity to be their own fleet managers. Diamandis’s appealing image—displaced drivers buying one or more cars that earn while they relax—drew a practical qualification from Khosrowshahi: “There’s always some work involved,” including maintaining the fleet.

• Diamandis cited roughly 1 million active US drivers and roughly 10 million globally, with Khosrowshahi correcting the US figure upward. Looking toward 2030, Khosrowshahi expects significantly more drivers on the platform. Uber is growing over 20%, its audience almost 20%, and around 20% of drivers leave each year; his approach is to absorb autonomous deployment by slowing recruitment so existing drivers retain comparable work.

• His more consequential hedge is that robotaxis might enlarge mobility demand. Austin and Atlanta are growing faster than Uber’s national average, suggesting autonomy attracts new customers; cheaper rides would strengthen that effect. Uber’s economic product, in his formulation, is ultimately simple: “give your time back.”