Originally published June 2017. Updated June 14, 2026.
Uber's autonomous vehicle program ran across nine years, absorbed the largest single trade-secrets litigation in modern technology industry history, produced the first pedestrian fatality involving a self-driving vehicle, and was ultimately sold to Aurora Innovation in 2020 — exiting Uber from direct autonomous vehicle operation entirely. The case is the most-studied AV operating record in the contemporary technology industry. The lessons run across acquihire risk, talent-recruitment ethics, safety operations under commercial pressure, regulatory engagement, and whether autonomous vehicle technology can scale on the timelines the post-2014 AV industry consistently projected.
The Otto acquisition — August 2016
Uber acquired Otto — the autonomous truck startup founded in early 2016 by former Google self-driving car project employee Anthony Levandowski — for approximately $680 million in August 2016. Levandowski was named head of Uber's broader autonomous vehicle operations. The acquisition was structured to give Uber rapid access to AV technology, talent, and intellectual property at scale.
The acquisition was the inflection moment that triggered the subsequent operational and legal crisis. Within four months of the Otto acquisition closing, Google's autonomous vehicle subsidiary Waymo filed a federal lawsuit alleging Levandowski had taken approximately 14,000 confidential files from Google in advance of his Otto founding — and that Uber had structurally encouraged the trade-secrets theft.
The Waymo v. Uber litigation
The February 2017 Waymo v. Uber complaint produced federal litigation across the next nine months. The trial proceeded across early 2018 in U.S. District Court for the Northern District of California. The litigation was settled on February 9, 2018 — five days into the trial — for approximately $245 million in Uber equity, equivalent to approximately 0.34% of the company.
Anthony Levandowski was subsequently indicted by federal prosecutors on 33 counts of trade-secrets theft. He pled guilty in March 2020 to a single count and was sentenced to 18 months in federal prison. President Trump pardoned Levandowski in January 2021 — making the case one of the most-cited single technology-industry federal pardons of the post-Trump period.
The Tempe, Arizona fatality — March 2018
On March 18, 2018, an Uber autonomous vehicle operating in autonomous mode struck and killed Elaine Herzberg as she crossed a street in Tempe, Arizona at approximately 10:00 PM. Herzberg was the first known pedestrian killed by a fully autonomous vehicle in modern transportation history.
The post-fatality investigation produced operational consequences. The National Transportation Safety Board's final report, issued in November 2019, concluded that the immediate cause was the vehicle's failure to properly classify the pedestrian. The contributing causes included inadequate Uber safety culture, inadequate driver monitoring of the autonomous system, and inadequate operational discipline across the broader AV program.
The vehicle's safety driver — Rafaela Vasquez — was charged with negligent homicide in 2020 and pled guilty to endangerment in 2023. Uber was not criminally charged. The civil settlement with Herzberg's family was reached for an undisclosed amount.
The Tempe fatality reset AV operational expectations across the entire industry. The cumulative effect was an operational discipline upgrade across Uber, Waymo, Cruise, and the broader competitive set. The fatality was the single most-consequential safety event in modern AV development.
The Aurora sale — December 2020
Uber announced the sale of its autonomous vehicle unit — Advanced Technologies Group (ATG) — to Aurora Innovation in December 2020. The transaction was structured as a stock-for-stock exchange. Uber received approximately 26% of Aurora and committed an additional $400 million investment. Aurora absorbed the ATG operations, the Pittsburgh-based engineering operation, and a substantial portion of the cumulative intellectual property.
The sale represented Uber's exit from direct autonomous vehicle operation. Dara Khosrowshahi — who had replaced Travis Kalanick as Uber CEO in September 2017 — had concluded that the AV development timeline did not justify the capital allocation required for direct development.
The post-2020 environment — Aurora, Waymo, Cruise, Tesla
The post-2020 AV competitive environment has restructured substantially. Aurora went public via SPAC in November 2021 and has continued autonomous trucking development. Waymo has expanded commercial robotaxi operations in Phoenix, San Francisco, Los Angeles, and Austin across 2024–2026. General Motors' Cruise division absorbed operational setbacks, suspended robotaxi operations following a 2023 pedestrian incident, and was substantially restructured across 2024. Tesla has continued Full Self-Driving development inside its promised-timeline pattern under Elon Musk's continued operational involvement — context covered in The Elon Musk Political Arc.
Uber has returned to partnership-driven AV integration rather than direct development. Uber partnerships with Waymo, Aurora, and additional autonomous vehicle operators have produced commercial autonomous ride and delivery service across multiple U.S. markets across 2024–2026.
Where this fits in the Uber crisis archive
The Self-Driving arc is one of the six major operating sub-arcs in Uber's twelve-year communications case archive. The Waymo lawsuit broke in the same February-March 2017 window as the Susan Fowler memo and the Greyball revelations — the three-week stretch that triggered the Holder investigation and ended in Travis Kalanick's resignation. The Tempe fatality landed in March 2018, a year after the Khosrowshahi-era reset began. The Aurora exit in December 2020 was the operational expression of the reputational reset that Uber's modern communications function was built to sustain. The complete sequence — surge pricing through Women Preferences — is mapped in EPR's Uber PR Crises Timeline 2014–2026. The regulatory parallel in Europe — the 2017 ECJ Uber Spain ruling, the 2021 UK Aslam decision — is broken out in Uber, Airbnb, and Europe. Ronn Torossian's 2014 column on Uber and the law diagnosed the underlying pattern before any of the named episodes broke.
The operating reads
Acquihire transactions carry legal exposure when the acquired talent's prior employer maintains strong intellectual property infrastructure. The Otto-Uber-Waymo arc demonstrated the pattern at significant scale.
Safety operations require operational discipline that commercial pressure compresses. The Tempe fatality NTSB report documented inadequate safety culture under commercial development pressure.
Autonomous vehicle commercial timelines have consistently extended beyond initial projections. The 2016–2020 AV industry projected commercial robotaxi operations within five years. The 2024–2026 commercial operating record is more limited than the 2016 projections anticipated. The pattern has held across Waymo, Aurora, Cruise, Tesla, and the broader competitive set.
Partnership-driven integration may be the sustainable commercial pattern. Uber's post-2020 partnership model with Waymo, Aurora, and additional autonomous operators represents the alternative to direct development. Whether the partnership model produces sustainable commercial economics is the open question of the next five years — the same kind of platform-versus-direct-operation question visible across Apple TV's content strategy.
The Levandowski pardon is the canonical case study in technology industry federal pardon dynamics. The pardon produced discussion of executive-branch pardon authority applied to technology industry crimes — a discussion that has continued across the post-2021 period.
The verdict
Uber's autonomous vehicle program is the most-studied direct AV operating record in modern technology industry history. The Otto acquisition, the Waymo litigation, the Levandowski criminal case and pardon, the Tempe fatality, and the Aurora exit together produced the canonical case study. The operational lessons are now embedded in the broader AV competitive set's operating discipline. The structural questions about AV commercial timelines remain unresolved across the 2024–2026 cycle. Uber's exit from direct AV operation was operationally correct in retrospect — the commercial timeline that would have justified sustained development did not materialize on the 2017 projected schedule.
Everything-PR Coverage of Uber
Crisis history
Women Preferences 2026
Marketing
Broader platform-arc context
