Uber just announced a $1.25 billion investment in Rivian to deploy up to 50,000 robotaxis through 2031, marking the ride-hailing giant’s return to autonomous vehicles after spectacularly failing the first time. In 2020, Uber sold its Advanced Technologies Group (ATG) to Aurora for a fraction of the billions it spent, after losing $303 million in just nine months. Now they’re back, but with a smarter playbook: partnership over internal development, milestone-based tranches instead of open-ended cash burn, and a 2028 target that’s either ambitious or delusional depending on who you ask.
From ATG Disaster to Calculated Risk
Uber’s self-driving journey reads like a cautionary tale. Travis Kalanick launched ATG in 2016 after poaching 40 researchers from Carnegie Mellon, convinced that owning autonomous technology was existential. By December 2020, after safety incidents and hundreds of millions in losses, Uber sold the 1,200-person division to Aurora for equity in a $10 billion valuation. Dara Khosrowshahi, who inherited the mess, got a 26% Aurora stake and a board seat, but no cash return on the billions burned.
The Rivian deal flips the script. Uber will invest $300 million upfront, with four additional tranches tied to undisclosed milestones through 2031. That structure caps downside risk while maintaining upside exposure, a stark contrast to ATG’s bottomless pit. Rivian gets a guaranteed customer for 10,000 R2 electric SUVs (with an option for 40,000 more) and a badly needed $1.25 billion cash injection. Uber gets an autonomous fleet without building the tech themselves.
The 2028 Timeline Problem
Here’s where skepticism is warranted: Uber promises robotaxis in San Francisco and Miami by 2028. That’s two years away. The Rivian R2 just launched this spring with a $45,000 starting price, 300-mile range, and what Rivian vaguely calls “autonomous capabilities.” Translation: 11 cameras and 5 radars are installed, but there’s no proven autonomous driving software stack. Rivian has exactly zero miles of real-world robotaxi testing.
Compare that to Waymo, which has spent over a decade developing its technology and now operates 1,000+ vehicles completing millions of rides monthly in San Francisco, Phoenix, and Los Angeles. Uber’s own ATG division worked for four years and couldn’t crack it. Now they’re betting Rivian, or some unnamed partner, can deliver production-ready autonomous R2s in 24 months?
The announcement conveniently omits who’s actually building the self-driving software. Is it Rivian? A third-party supplier? Some stealth partnership yet to be disclosed? Without that answer, the 2028 timeline looks more like investor relations theater than engineering reality.
Robotaxi Market: Multi-Billion Dollar Battleground
Uber’s urgency makes sense given the market dynamics. The robotaxi sector is projected to grow from $5.5 billion in 2026 to $147 billion by 2033, a 99% compound annual growth rate. San Francisco is becoming a four-way battleground: Waymo dominates today, but Amazon’s Zoox is launching late 2026, Tesla looms as the “biggest potential threat,” and Uber itself plans a separate Lucid-powered robotaxi service (yes, two different robotaxi partnerships) also launching in late 2026.
The Rivian partnership offers strategic advantages over buying Waymo rides wholesale or partnering with Tesla. Rivian has no competing ride-hailing ambitions, gives Uber platform exclusivity, and allows co-branding around the premium EV experience. It’s a cleaner relationship than dealing with Elon Musk or bending the knee to Alphabet.
Gig Workers Face the 15-Year Countdown
While Uber pitches robotaxis as the future, current drivers are watching their livelihoods erode in real time. In the five metros where robotaxis currently operate, drivers completed 5.3% fewer trips per hour in Q4 2025 compared to the year before. That’s not a projection, it’s already happening.
Uber CEO Dara Khosrowshahi has been candid about the endgame: autonomous vehicles will replace most human drivers within 15 years. The company is preparing “alternative opportunities” like AI data labeling and robotaxi fleet maintenance, but those roles pay less and employ fewer people than driving. Goldman Sachs projects 35,000 robotaxis capturing 8% of the U.S. ride-share market by 2030, with S&P Global forecasting parity with human drivers by 2041.
The economic impact extends beyond drivers. A Waymo ride sends revenue to Alphabet’s Mountain View headquarters, with minimal local economic circulation beyond fleet maintenance. Cities welcoming robotaxis may discover they’ve invited a service that extracts rider spending without creating proportional local jobs.
Smart Pivot or Déjà Vu?
Uber learned expensive lessons from ATG: building autonomous tech internally is harder and costlier than ride-hailing executives imagine. The milestone-based Rivian deal shows they absorbed that lesson. But the aggressive 2028 timeline, vague technology partnerships, and absence of any proven Rivian autonomous capability raise familiar questions. Is this the disciplined return of a chastened company, or ATG 2.0 with better PR?
We’ll know by 2028. Either Uber launches robotaxis in San Francisco as promised, or they issue another press release explaining why partnerships are hard and technology takes time.
