Uber CEO Dara Khosrowshahi on self-driving's future, changing business model, job displacement

Watch on YouTube ↗  |  September 17, 2025 at 21:22  |  25:37  |  All-In Podcast
Speakers
Dara Khosrowshahi — CEO, Uber
Jason Calacanis — Angel Investor / Founder, LAUNCH

Summary

Uber CEO Dara Khosrowshahi discusses Uber's autonomous-vehicle strategy with the All-In hosts: 20-plus AV partners including Waymo and China's Baidu, WeRide and Pony.ai, Tesla's camera-only approach versus sensor-heavy stacks, and why Uber's hybrid human-plus-robot network should keep a utilization moat as robotaxis scale. He argues autonomy expands the mobility market, that robotaxi fleets will eventually be owned by REIT-like financial players while Uber takes only transitional balance-sheet risk, and that $8.5B of trailing cash flow funds both aggressive AV investment and the $20B buyback. The conversation also covers eVTOL (Joby), sidewalk robots (Serve) and drone delivery, the on-demand delivery imperative for food and retail, and the long-run driver job-displacement question.

  • Uber has over 20 autonomy partners; Waymo is live on Uber in Austin and Atlanta, and Chinese L4 operators Baidu, WeRide and Pony.ai are partners expanding outside China.
  • Khosrowshahi contrasts Tesla's camera-only, no-HD-map, lean-compute approach with sensor-redundant, HD-mapped stacks, calling himself an interested bystander.
  • He argues Uber's demand density gives AV fleets far higher utilization and revenue per car than going direct, so even Waymo or Tesla would benefit from the network.
  • End state: REIT-like financial owners hold robotaxi fleets; Uber takes balance-sheet risk only to prove per-car economics, then financializes.
  • Uber is an investor in Joby (eVTOL) and works with Serve and others on sidewalk robots; robots plus drones could cover 50%+ of delivery TAM.
  • $8.5B trailing cash flow with 18% top-line and 35% bottom-line growth funds both aggressive AV investment and the $20B buyback.
  • Food, grocery and retail players must adopt on-demand delivery or lose share; robotization should lower food costs and lift Uber's delivery business.
  • Robotaxis should not displace drivers for 5-7 years given growth and turnover, but displacement becomes a real issue in 10-15 years.
Ideas
Chinese L4 players are strong, expanding abroad.
Autonomy in China is hitting the big time: Baidu, WeRide and Pony.ai are already on the road with no safety driver, their capabilities are amazing given how complex driving in big Chinese cities is, they take safety as seriously as Western companies, and their safety record is excellent. The Chinese players that want to expand outside China are partnering with Uber, which works with all of them, giving them distribution as they go international.
Autonomy greatly expands Uber's mobility market.
Uber's marketplace economics survive the shift to robotaxis. A standalone fleet has fewer vehicles and no demand of its own, while Uber's hybrid human-plus-autonomous network already has the demand and sends much closer pickups (3 minutes instead of 15 for a 10-minute ride), so revenue-generating miles as a share of total miles and revenue per car per day are much higher for a fleet on Uber than for one going direct. Just as McDonald's runs a direct channel and still sells through marketplaces, even Waymo or Tesla will want to work with Uber to maximize the economics of very expensive cars. The end state is financial, REIT-like owners holding big fleets on Uber's network; Uber will use its balance sheet only to prove the revenue per car and then move fleets off balance sheet, staying largely capital-light as the demand layer for asset-heavy operators.
Uber backs Joby for eVTOL transport.
Uber is an absolute believer in eVTOL and is an investor in Joby; it will work with Joby and put its aircraft on the network as the vehicles become available. Cities have expanded into the third dimension for businesses and residences, but transportation infrastructure has only grown in two dimensions, which is why traffic keeps getting worse, so the 'Z axis' makes a ton of sense and Uber's demand can feed it.
Uber scales sidewalk-robot delivery with Serve.
Sidewalk delivery robots are easier tech to develop, drive slowly, are very safe and suit deliveries of a mile or less in dense areas (already running in LA and Santa Monica), so they address a specific slice of Uber's delivery market; Uber is working with Serve Robotics and a number of other players in the US, Japan and other markets. Together with drone delivery for spread-out suburban areas, robots could cover 50%+ of Uber's delivery TAM, with the first and last mile in and out of buildings the remaining challenge.
Up Next

This All-In Podcast video, published September 17, 2025, features Dara Khosrowshahi discussing BAIDU, WeRide, PONY, UBER, JOBY, SERV. 4 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Dara Khosrowshahi  · Tickers: BAIDU, WeRide, PONY, UBER, JOBY, SERV