[Robotaxi Wars Among Korea, China, and the US] ④ The Race of the Driverless Box... The 'Final 1%' Data Battle Led by US Big Tech

Interior photo of Zooks vehicle Photo by Paek Jongmin, Tech Specialist

At the Consumer Electronics Show (CES) held in Las Vegas, USA this past January, what drew the most attention from attendees was not the exhibition hall itself, but the autonomous robotaxis.

Those who downloaded the app on their smartphones and made reservations to experience the autonomous robotaxi by Amazon subsidiary Zoox were dismayed to see a waiting time of over four hours, yet still waited for their turn to ride.

The reporter also joined the line. Perhaps luckily, the initial waiting time displayed was 40 minutes, but soon after, a notification popped up on the phone stating the vehicle would arrive in 20 minutes.

A signal came that the vehicle had arrived in front of the hotel, the starting point. The vehicle did not resemble a typical sedan, SUV, or bus. It was box-shaped, with the front and rear looking identical. In other words, there was no distinction between the front and back. The interior was even more unconventional. There was no driver's seat. Naturally, there was no steering wheel or brake pedal.

Sitting in seats arranged to face each other in pairs, it felt like riding a Ferris wheel or a small cable car at an amusement park.

Once the doors closed, the vehicle began to move quietly. Since it was nighttime, ambient lighting turned on inside, and Amazon's music service automatically played background music. Although there was no driver, the vehicle stopped at signals and navigated intersections while avoiding pedestrians.

The driving speed was fast as well. The assumption that an autonomous taxi would move slowly turned out to be wrong. It felt like the work of a veteran driver. The route was displayed on a screen inside the vehicle. Upon arrival at the destination, an announcement said the vehicle was searching for a safe parking spot. After completing this impressive experience, the vehicle headed off to its next passenger. Zoox is currently operating robotaxi services in both Las Vegas and San Francisco.

Unmanned Mobility Has Become Routine... Waymo Dominates San Francisco

In downtown San Francisco, robotaxis have effectively taken over the taxi market. In November last year, the reporter often saw taxis equipped with various sensors around Pier 39, a popular tourist attraction in San Francisco.

On the bustling roads shared by tourists, regular vehicles, and trams, these unusual-looking taxis with multiple sensors continuously picked up and dropped off passengers. These were Waymo's robotaxis, a subsidiary of Google.

Unlike a few years ago when the reporter last visited San Francisco, the number of Waymo vehicles has surged. Waymo vehicles have recently started operating on highways as well.

Jux Robotaxi opened its door to welcome passengers. Photo by Paek Jongmin, Tech Specialist

Until just a few years ago, most of the taxis in the area were the yellow cabs or Uber vehicles. Now, the scene has changed completely. According to the city of San Francisco, regular taxis have been virtually driven out of the area. There are currently about 1,000 Waymo taxis in operation just within the San Francisco region. A structure has formed in which ride-sharing taxi services like Uber and Lyft, as well as Waymo, share responsibility for urban mobility.

The fact that robotaxis have become part of urban transportation was also evident in an unexpected event. This past January, when a large-scale power outage occurred in San Francisco, Waymo cars all came to a halt simultaneously. While the confusion among passengers exposed the blind spots of autonomous driving, it also illustrated how robotaxis have now become a critical part of the city’s transportation system.

Google Waymo robotaxi is driving alongside regular vehicles and trams around Pier 39, a tourist attraction in San Francisco. Photo by Paek Jongmin, Tech Specialist

Tesla’s Ambitious Plans to Catch Up With Waymo... 'Cybercab' Mass Production Nears

Tesla is considered a dark horse in the U.S. robotaxi market, despite being behind the competition. While Tesla already offers its Full Self-Driving (FSD) service, CEO Elon Musk’s ambitions extend further. His goal is the two-seater robotaxi called "Cybercab." Unveiled in 2024, the Cybercab aims for full-scale mass production this year. Of course, many remain skeptical about whether mass production will actually begin as promised.

In June of last year, Tesla drew attention by launching a pilot robotaxi service using Model Y vehicles in Austin, Texas. This served as a warm-up before the full debut of the Cybercab. In this service, the driver’s seat is empty, with only a supervisor seated in the passenger seat.

Despite the limited scope of the service, bloggers and YouTubers have shared firsthand accounts of their rides. Kim Junbum, CEO of HL Holdings, who tried the Tesla robotaxi while on a business trip to Austin, commented, “I was amazed by how well it drove.”

Since the beginning of this year, Tesla’s robotaxi has also begun introducing fully unmanned operations without safety officers in stages.

Tesla's Robotaxi 'Cybercap' Reuters Yonhap News

Three Companies, Three Strategies... The Dawn of Competition in Sensors, Vision, and Platforms

The current U.S. robotaxi market is a three-way race among Waymo, Tesla, and Zoox. While the consensus is that Waymo is currently a step ahead in terms of commercialization, some argue that the competition has only just begun.

Each company's approach to robotaxis stands in stark contrast throughout this process. Both Zoox and Waymo vehicles are equipped with a variety of sensors, including LiDAR, radar, and cameras, to perceive their surroundings. Their goal is safe driving through a combination of vision and sensors. In Tesla's case, under Musk’s direction, the company shifted to a vision-only approach. Musk was inspired by the idea that humans drive using only their eyes and thus transitioned to autonomous driving using only camera footage.

Waymo’s greatest competitive edge lies in its data. Its accumulated autonomous driving distance is reportedly over 160 million kilometers, making it the autonomous driving company with the largest amount of real-world road data in the world.

Tesla is targeting the robotaxi market with another strategy. While Tesla’s robotaxi fleet is absolutely smaller than Waymo’s, its advantage is the massive amount of data collected from its existing vehicles already sold.

At CES 2026, Musk commented on Nvidia’s autonomous driving AI platform "Alpamayo," stating, “Anyone can achieve 99%, but only Tesla can solve the remaining 1%.” This means that while autonomous driving technology can handle most situations, the final stage—dealing with unforeseen or unpredictable circumstances—is the most difficult. Musk’s remarks express his confidence that Tesla has an edge at this critical point.

The vehicles themselves also differ. Zoox and Tesla both use dedicated vehicles. Because Zoox manufactures box-shaped dedicated vehicles, costs are high. In contrast, Tesla has taken a strategy of evolving into a two-seater Cybercab, which closely resembles existing vehicles but lacks a steering wheel and pedals.

As a Google subsidiary, Waymo has stepped away from focusing on the vehicle itself. Its core is self-driving software—in other words, a self-driving software platform strategy. Furthermore, Google’s map service provides the foundation for vehicle navigation. Waymo utilizes existing electric vehicles such as the Jaguar I-PACE. In the future, Hyundai’s IONIQ 5, produced in the U.S., and Zeekr vehicles from China will also be used as robotaxis. These vehicles were also displayed at CES 2026. It’s similar to how Google supplies only the operating system without manufacturing Android smartphones.

Ultimately, It’s an AI Infrastructure War... Big Tech Builds Massive Barriers to Entry

What deserves particular attention is that the robotaxi competition is not simply a contest of vehicle technology, but is built upon a background of AI computing infrastructure competition. Major tech companies such as Amazon, Tesla, and Google are training autonomous driving AI on data centers equipped with massive numbers of graphics processing units (GPUs) for AI learning and inference. Tesla has built its own supercomputer, Dojo, to train its autonomous driving AI, and is expanding its data centers while forming an alliance with SpaceX and xAI.

Google is also leveraging its AI infrastructure and data centers for Waymo’s development. Google’s Gemini AI may also serve as a strong ally for Waymo. Amazon supports Zoox’s AI training through AWS, the world’s leading cloud service.

This is why it is believed that these three companies are building formidable barriers to entry that make it extremely difficult for other challengers to catch up.

This content was produced with the assistance of AI translation services.