Baidu Apollo Go Tops Waymo in New Global Robotaxi Index

Comparing US and Chinese autonomous vehicle programs has largely been a matter of assertion, with each side citing different metrics. A new benchmarking effort from Autnmy AI, the Road to Autonomy Index, attempts something more systematic: a real-time public scorecard built from global public databases, SEC filings, and federal and state reports, updated every 12 hours. According to TechCrunch, the index currently ranks Baidu Apollo Go ahead of Waymo in the robotaxi category, with Pony.ai, WeRide, and Tesla completing the top five. Three of those five slots go to Chinese companies.

The platform scores companies across six factors, each graded from 0 to 100: autonomous operations, scale, revenue, commercial partnerships, manufacturing, and safety transparency. The first three account for 70 percent of each company's composite score. Co-founder Rob Grant described the sourcing: "We don't scrape information...if it's publicly available or under Creative Commons license, we use it." The index covers four vehicle segments: robotaxis, autonomous driving licensing, autonomous trucks, and delivery robots. The methodology is explicit that it aims to "reward commercialization, not promises," with the algorithm heavily weighted toward fully autonomous operations generating revenue rather than demos or pilots still running with safety drivers.

The caveats matter. An index built substantially on SEC filings and US federal disclosures may not fully capture companies whose primary operations and regulators are in China, and the 12-hour refresh cycle means any snapshot is a momentary reading rather than a settled verdict. What the reporting does not address is how Autnmy AI accounts for cross-market disclosure differences, or what its own business model looks like.

For investors and policymakers tracking the global AV race, three Chinese companies appearing in the top five of a data-sourced public benchmark is a more concrete data point than the usual anecdotal comparisons, even accounting for the methodology uncertainty that a first-generation index carries.