An AI that re-ranks the robotaxi race every 12 hours just put Tesla in fifth, and the top of the list is mostly Chinese.
A new AI-built benchmark called the Road to Autonomy Index now ranks the world’s robotaxi players in near real time, and the early results flip the usual story. As of Friday, China’s Baidu Apollo Go sat at number one, Waymo second, Pony.ai and WeRide right behind, and Tesla down in fifth.
Picture the last decade of self-driving hype: endless flashy demos, mountains of capital, and a parlor game about who was “winning” that nobody could actually measure. There was no scoreboard, just vibes and press releases. Now there’s a scoreboard. And it does not say what most of Silicon Valley assumed it would.
What Happened
Advisory and research startup Autnmy AI built a generative AI platform to benchmark and rank autonomous vehicle companies, then this week released the Road to Autonomy Index to the public. The system pulls from public databases, including federal and state reports, SEC filings, and public exchanges, then weighs each company on operations, scale, revenue, commercial partnerships, manufacturing, and safety record. It refreshes the standings every 12 hours.
The headline result: the robotaxi leader was not Waymo. It was Baidu’s Apollo Go program, just barely. Waymo landed second, followed by fellow Chinese players Pony.ai and WeRide, with Tesla in fifth.
The Backstory
For ten years the self-driving conversation has been an American one. Waymo, Tesla, Cruise, the usual cast. The whole framing quietly assumed the race would be won in California or Austin.
The problem was always measurement. As Autnmy AI’s reporting puts it, the early autonomy era was rich in demos and capital but thin on substance the public could actually verify. The Index is an attempt to fix that with data instead of narrative. It runs four separate rankings covering robotaxis, autonomous-driving licensing companies, autonomous trucks, and delivery bots.
Co-founder Rob Grant is also careful about how the data gets sourced. The platform does not scrape the web, he says. It uses information that is publicly available or under a Creative Commons license, plus some data it pays to license.
The Plan
Take a global view of the data, and the picture changes. Grant said one of the takeaways that stuck with him early was how strongly Chinese companies ranked across multiple categories, not just robotaxis.
It is a useful gut check against the home-team bias baked into US tech coverage. A benchmark that weighs scale, revenue, and commercial operations rewards the companies actually moving people around at volume, and on those terms, Apollo Go, Pony.ai, and WeRide stack up. The scoreboard is not measuring who has the best demo. It is measuring who is running a business.
Meanwhile, on the US side, the ground game is still very real. Texas launched an automated-vehicle tracker in May, and the fleet counts show who is building. As of late May, Waymo had 577 autonomous vehicles registered in the state. It now has 620, roughly a 7.5% jump in under a month. Tesla went from 42 to 69, a 64% increase. Zoox climbed from 35 to 43. Avride, Nuro, and Volkswagen subsidiary MOIA held steady at 317, 47, and 12.
The Business Model Angle
Here is the pattern worth stealing. When a market has no agreed way to keep score, the loudest player tends to “win” by default, because perception fills the vacuum. Build a credible measurement layer, and suddenly the standings reflect operations instead of marketing budgets.
That is the same lesson lurking under the robotaxi economics themselves. As we broke down in our look at why self-driving AI is only 7% of a ride’s cost, the flashy autonomy layer gets the headlines while the unglamorous operational layer is where the durable money sits. A scorecard that weights scale, revenue, and real commercial deployment is, in effect, scoring the boring layer. And the boring layer is exactly where Chinese operators have been quietly racking up rides.
For founders, the takeaway is blunt: in a hyped category, the company that defines the metric shapes the market. Autnmy AI did not build a robotaxi. It built the referee. That is a business too.
The Risk
A few honest caveats. This is a brand-new index from a startup with a clear interest in being seen as the authoritative scorekeeper, and “Baidu leads, just barely” is a margin thin enough to flip on the next 12-hour refresh. A ranking that re-sorts twice a day is great for engagement and tricky for drawing firm conclusions.
Cross-border comparisons are also messy. China and the US run different regulatory regimes, different city densities, and very different definitions of a deployed robotaxi, so weighting “scale” globally can quietly favor whoever operates in the most permissive markets. And US momentum has its own asterisks: Zoox still cannot charge for rides until it gets a federal exemption, and Waymo recently recalled nearly 4,000 robotaxis to stop them from driving into highway construction zones, with the software fix still under development. A scoreboard is only as good as the data feeding it, and the data here is moving fast.
Quick Questions
Who is actually winning the robotaxi race right now?
By the Road to Autonomy Index, China’s Baidu Apollo Go ranked first as of Friday, narrowly ahead of Waymo, then Pony.ai and WeRide, with Tesla fifth. The margin at the top is slim.
What is the Road to Autonomy Index?
It is an AI-built benchmark from startup Autnmy AI that ranks autonomous vehicle companies using public data on operations, scale, revenue, partnerships, manufacturing, and safety, updated every 12 hours across four categories.
Why is China ranking so high if US robotaxis get all the attention?
The Index takes a global view and weighs real commercial operations and scale, not demos or media coverage. On those terms, Chinese operators like Apollo Go, Pony.ai, and WeRide stack up strongly.
Where does Tesla land?
Fifth on the robotaxi index as of Friday. On the ground in Texas, though, Tesla’s registered autonomous fleet grew fast, from 42 to 69 vehicles, a 64% jump in under a month.
The Business Model Analyst Take
The robotaxi race just got a referee, and the first whistle went against the home crowd. Whether or not Baidu stays on top through the next refresh matters less than the bigger shift: the industry now has a measurement layer, and measurement layers reshape narratives. The smart move for operators is not to argue with the scoreboard. It is to notice that someone just built one, monetized being the scorekeeper, and quietly reminded everyone that the company defining “winning” often wins more than the companies playing.
Based on reporting by Kirsten Korosec for TechCrunch.
