Uber and Nvidia set 2027 launch for self-driving taxi service

Uber (UBER) is teaming up with Nvidia (NVDA) to launch self-driving taxis in Los Angeles and San Francisco in 2027.

That’s a scheduled rollout in two of the most important urban markets in the U.S., at a time when other autonomous robotaxi strategies are also starting to take shape.

Uber plans to deploy 100,000 vehicles, such as the Lucid Gravity, powered by Nvidia’s Drive Hyperion platform. At the same time, Nvidia is making a push to become the backbone for robotaxi systems worldwide.

The companies are building a network that could expand to 28 cities across four continents. That’s the early stages of a global rollout.

Uber and Nvidia just put a firm timeline on autonomy, centered around Level 4 systems. That means vehicles can handle driving on their own in specific areas without human input. Nvidia supplies the full technology stack, including compute, sensors, and its Alpamayo 1.5 model, which can train for rare driving situations using plain language instead of manual coding.

Uber brings the other half of the equation. It already controls demand through its rider base, routing, and pricing systems, plus global market access.

Lyft, Bolt, and Grab are also adopting Nvidia’s platform. That positions Nvidia as a core infrastructure layer across multiple ride-hailing companies.

The disruption behind the news: This is a platform war, not a taxi story.

Nvidia is becoming the Intel of autonomy, and Uber is becoming the AWS of mobility.

Everyone else is either a fleet operator or gets squeezed out.

Whoever owns the autonomy stack controls cost per mile. Today, human drivers cost Uber roughly 60 percent to 70 percent of every ride. Remove that, and the unit economics flip fast. A robotaxi running 20 hours a day can generate 2 times to 3 times the revenue of a human-driven car, with lower operating costs over time. That’s what drives adoption.

And Nvidia is making itself hard to avoid. If Uber, Lyft, Bolt, and Grab all run on the same core stack, Nvidia captures upside across the entire category, not just one platform. That’s how chips scale into a multitrillion-dollar robotics business.

The second shift is speed. Alpamayo 1.5 reduces training friction. Instead of manually coding rare driving scenarios, developers can describe them. That compresses development cycles from months to days. It leads to faster iteration, faster safety improvements, and faster regulatory approval. That’s how you scale from 2 cities to 28.

Competition is already reacting. Waymo is expanding its robotaxi footprint with newer-generation vehicles in the U.S. Zoox is adding markets, and Amazon is preparing to use Uber’s platform to scale Zoox robotaxis across the country. But here’s the difference. Uber doesn’t need to win the tech race. It just needs to aggregate supply. If Nvidia wins the stack, Uber controls the demand layer.

This creates pressure across the market. Smaller autonomy startups can’t compete on training data or deployment scale. Traditional automakers lose pricing power as software takes more of the value. Even city regulators will be forced to adapt as fleets scale faster than policy cycles.

What to watch next

Watch cost per mile drop below $1.

Watch fleet utilization rates climb past 70 percent.

Watch human drivers get pushed out of dense urban routes first.

The next 24 months will focus on positioning. Partnerships, city agreements, and regulatory groundwork. By 2027, the switch flips. The companies that locked in infrastructure early will dominate. The ones still testing will fall behind.

Also watch Nvidia’s position strengthen. If multiple ride platforms standardize on Drive Hyperion, switching costs rise sharply. No one will want to rebuild their autonomy stack from scratch. That’s how Nvidia shifts from supplier to gatekeeper.

What’s easy to miss here is pricing power.

If autonomy brings the cost per mile down to about $0.80, but rides still cost $1.50 to $2.00, that extra margin doesn’t go back to riders right away. It stays with the platform that controls demand.

At Uber’s scale, we’re talking tens of billions of trips a year. So even a $0.50 bump in margin per ride adds up to tens of billions in extra gross profit. That gives them a lot of flexibility. They can cut prices to squeeze competitors or spend aggressively to pull in more riders.

Nvidia (NVDA) and Uber (UBER) have a Disruption Score of 4 and 2, respectively. Click here to learn how we calculate the Disruption Score.  

Nvidia is also part of the Disruption Aristocrats, our quarterly list of the world’s top disruptive stocks.