Robotaxis hit the streets: how Waymo, Tesla and Baidu are fighting glitches

Robotaxis have stopped being the perpetual "coming soon." Waymo, Baidu and other companies are already carrying real passengers on the streets of American and Chinese cities. The technology has finally reached the road, albeit with caveats.

What already works

Waymo is the undisputed leader in the US. The company completes more than 100,000 trips a week in San Francisco, Phoenix and Los Angeles with no driver behind the wheel. Users order a car through an app and ride while the lidar sensors and cameras on the roof do the rest.

Baidu's Apollo Go works on a similar model in China. In Wuhan alone, hundreds of vehicles carry passengers every day, and some rides already have no operator in the car. The service is gradually spreading to major cities: Shenzhen, Chongqing and Beijing.

The ecosystem is expanding through partnerships: Uber and Lyft have built partners' robotaxis into their apps, turning driverless transport into an option inside familiar services.

Key market figures:

• Waymo has driven more than 50 million miles in autonomous mode
• Baidu Apollo Go has completed over 8 million commercial rides
• Tesla plans the Cybercab, a two-seat electric car without pedals, by the end of 2026
• Amazon's Zoox is testing a bidirectional robotaxi van in San Francisco

Why glitches persist

The industry still has systemic problems. Robotic cars handle non-standard situations worse: dense fog, downpours, a dug-up road, a pedestrian darting across on a red light. In such scenarios a car can freeze in the middle of an intersection and wait for instructions from a remote operator, which ruins the rider experience and creates traffic jams.

There have been reputational blows too. In late 2023, GM Cruise lost its license in California after one of its vehicles dragged a struck pedestrian several meters instead of stopping immediately. The incident set back trust in the industry by months. Waymo looks more careful: it publishes detailed safety statistics and expands its zones gradually.

The regulatory environment adds complexity. Some states require a separate license for each route, and in some cities authorities insist on an operator in the car, which wipes out the main savings from automation.

"The technology really works.

The question now is how to make it economically viable at scale," said a Bloomberg source close to one of the operators.

The race of business models

Tesla is taking a fundamentally different path. The Cybercab is a purpose-built electric car with no steering wheel or pedals, seating two passengers. Elon Musk names a target price of $0.20 per mile, several times cheaper than Waymo today. Skeptics doubt the timeline and the company's ability to scale production while also developing its autonomous-driving software stack, but if the Cybercab ships in the stated volumes, the market picture will change.

Waymo has raised several billion dollars from Alphabet, Toyota and sovereign funds and continues to invest in safety and coverage. Baidu is betting on government partnerships and subsidized fares, a strategy of capturing the market before it turns profitable that has already worked in other Chinese industries. Both companies expect scale to gradually make the economics converge.

What this means

Robotaxis have moved from the "eternal prototype" stage to "works, but not everywhere." The next two to three years will show which players can turn technological leadership into a sustainable business, and avoid stumbling over another regulatory ban or high-profile incident.