Uber just made its biggest bet on the future of transportation — and it doesn’t involve a single human driver.
At Nvidia’s GTC conference this week, Uber and Nvidia announced plans to deploy a fleet of 100,000 Level 4 autonomous robotaxis across 28 cities worldwide by 2028. The stock surged 5.6% on the news, hitting a multi-year high of $78.83. And honestly? The market might be underreacting.
The centerpiece of the deal is Nvidia’s new “Alpamayo” AI model — what Jensen Huang called the “ChatGPT moment for physical AI.” Unlike older autonomous systems that rely on pattern recognition, Alpamayo uses reasoning-based logic to navigate unpredictable real-world chaos: erratic pedestrians, surprise construction zones, the kind of stuff that makes self-driving hard. It’s a genuine leap in capability.
The timeline is aggressive but credible. Pilots are already running in Phoenix and Abu Dhabi. Full commercial launches start in Los Angeles and San Francisco in early 2027. A Tokyo pilot using Nissan vehicles is slated for late 2026. By 2028, the goal is 28 cities spanning North America, Europe, Asia, and Australia.
What makes Uber’s approach so clever is the asset-light model. Instead of building and maintaining its own fleet (the capital-intensive path Tesla and others are pursuing), Uber has locked in commitments from BYD, Geely, and Stellantis to manufacture “robotaxi-ready” vehicles pre-integrated with Nvidia’s hardware. Uber becomes the operating system for autonomous mobility without owning a single car.
The losers are becoming clear. Lyft, with its smaller R&D budget and no comparable global AI partner, is increasingly looking like a regional footnote. Traditional rental companies like Hertz and Avis are scrambling to reinvent themselves as fleet maintenance providers. And Tesla’s single-brand Cybercab approach, while technically impressive, can’t match Uber’s manufacturer-agnostic flexibility across dozens of markets.
The regulatory tailwinds are real, too. NHTSA proposed allowing mass production of vehicles without steering wheels or pedals earlier this year. The UK and EU have already harmonized autonomous vehicle standards. The question has shifted from “can it work?” to “how fast can it scale?” Uber’s answer: 100,000 cars in 28 cities in two years. That’s not a pilot program. That’s an invasion.
