Tesla Robotaxi Goes Live, Musk’s 12-Year-Old Promise Fulfilled

In the early hours of this morning Beijing time, Tesla officially launched its Robotaxi service on its official webpage.

According to the official description, Robotaxi is currently operating in a limited rollout only in the Austin, Texas. What’s confirmed, however, is that users can freely place orders via the app—the restrictions apply only to the operational area and the number of available vehicles.

As described by Tesla’s official AI account, the advantages of Tesla Robotaxi are as follows:

“Our Approach to autonomy also enables scalability: Tesla self-driving can be deployed anywhere it’s approved. It does not require expensive, specialized equipment or extensive mapping of service areas. It just works.”

Behind this confident and concise marketing copy is the milestone of “the world’s first vision-only Robotaxi trial operation.” Previously, all commercially operated Robotaxis, regardless of country, utilized LiDAR.

By the end of June 2025, a full 12 years have passed since Musk first promised autonomous driving services. Has he truly delivered? What’s the Tesla Robotaxi experience like? And has Tesla genuinely taken the next step as an AI company?

Hailing a Driverless Tesla

Based on shared videos from users on X, all available Robotaxis are Model Ys.

To ride in a Model Y with a steering wheel but no driver, you first need to be a member of Tesla’s “Early Access” program.

You’ll then receive a notification that the Robotaxi app is available for download. Here’s what it looks like—very cyberpunk:

Next, you need to confirm you’re within Robotaxi’s operational zone. Currently, service is mainly available within downtown Austin. Once your location is confirmed, you can hail a ride.

As a brand-new app, Robotaxi’s prompts are fairly detailed, and the order page includes relatively comprehensive usage reminders.

Additionally, after passengers board, the rear display enters Robotaxi reminder mode. Press and hold the large circle shown below to start the trip. There’s a dedicated UI, and functions like air conditioning and entertainment can also be controlled separately via the mobile app.

Currently, Robotaxi charges a flat rate of $4.20 per trip (approximately 29 RMB; the $4.20 figure—retaining two decimal places—is a well-known cultural reference). Tipping is temporarily unavailable.

Although a tipping option is displayed like Uber’s, tapping it reveals an official “Just kidding” notice. Perhaps the $12,000 FSD subscription fee has already factored into the “salary” for Hardware 4.0.

So, what’s the user experience like?

Since the number of Robotaxis deployed is currently low—estimated between 10-20 vehicles, though X users have spotted around 35 in service:

Overall, there aren’t many public reviews yet, and they mostly cover similar road conditions.

Additionally, likely for the same reason, there haven’t been any major negative posts so far.

In fact, since the rollout of FSD V13, U.S. autonomous driving experience videos have shown that, despite occasional red-light running, wrong-way driving, or navigation errors, the actual ride feel is remarkably stable, especially with industry-leading latency.

However, based on user feedback videos, even Robotaxi makes mistakes in certain situations. For example, in a video shared by X user @Rob Maurer, the system exhibited noticeable steering wheel jitter during a 22-minute ride.

It’s safe to say that even within the limited Austin downtown area, Tesla’s Robotaxi still has some flaws.

Coupled with the minuscule deployment scale, it’s clear Tesla isn’t yet competing head-to-head with today’s Waymo—which had over 1,500 vehicles across the U.S. and handled over 150,000 weekly rides by the first half of this year.

From the Beginning to a New Beginning

But before reaching today’s trial operation, Musk—representing the vision-only + automaker-driven autonomous driving approach—had his promises span three generations of Autopilot hardware and software.

In December 2012, after Tesla won its first major automotive magazine award, Musk told Motor Trend: “I think developing an automatic car would be really interesting.”

Autopilot launched in 2013, followed by the much-discussed partnerships and breakups with Mobileye and NVIDIA. Then came the game-changing FSD chip in 2019, and Hardware 4.0—still among the industry’s top computing platforms.

Throughout the history of Autopilot development, Musk’s obsession with “full self-driving” has been the constant thread.

In August 2013, he told CNBC: “Autonomous driving is definitely Tesla’s future”—back when Google hadn’t even invested in Uber, and Waymo hadn’t been named.

In late 2015, he told Fortune: “Within three years, Tesla cars will be able to drive themselves.” Musk first announced the “Robotaxi” plan at the 2019 Model Y launch. FSD Beta began rolling out in late 2020…

This journey included countless “two weeks” delays. The “Beta” tag on FSD persisted for three years. And even the stunning FSD V13 wasn’t quite the “hands-off-the-wheel Tesla” everyone envisioned.

Over these 12 years, Waymo stepped into the spotlight, Uber and DiDi launched and shelved autonomous projects, and Apple Car went through its entire “from inception to abandonment” cycle.

There are many angles to interpret Tesla’s 12-year journey. Today, we focus on this: All the immaturity, flaws, leadership, and progress of FSD are built upon Tesla’s near-stubborn insistence on a vision-only approach.

We don’t know if Tesla has hit the limits of vision-only, because despite delays, Musk and his team have steadily overcome its toughest challenges.

In the early hours Beijing time, Tesla’s AI team pressed the launch button. AI lead Srihari Sampathkumar called today: “A decade of hard work on to building best AI software and hardware to start Robotaxi service for the public. Today establishes the new era of physical AI dominance for Tesla”

A note on two key descriptors:

Around November 2015, tech media reported Musk had “defined autonomy as Tesla’s top priority.” During the 2020 earnings call and shareholder meeting, Musk described Tesla as a “leader in real-world AI” and “more like an AI and robotics company.”

Tesla AI senior engineer Phil Duan described today as: “It’s just a small step today. In hindsight, it will be the inflection point.”

Literally starting from zero, from vision-only, from an automaker’s perspective—this is the main thread of Tesla’s Robotaxi story. It’s just the opening chapter, far from proven success, but its greater significance lies in the step toward commercialization.

That said, we can conclude this: The Robotaxi trial marks the beginning of a new phase for Tesla, and indeed for the vision-only autonomous driving approach.

And as autonomous driving professor Philip Koopman of the Carnegie Mellon University told Reuters today, this trial is more like: “The end of the beginning – not the beginning of the end.”

Before Musk’s promise is fully realized and Tesla’s AI team celebrates its next victory, many challenges still lie ahead.