On October 24, 2026, at the World Intelligent Connected Vehicles Conference in Beijing, the Ministry of Industry and Information Technology released a set of figures that circulated quickly. As of August 2026, approximately 61 million passenger cars in China were equipped with L2 combined driving assistance. Of those, about 15 million had navigation-assisted driving — NOA, the feature that handles lane changes, merges, and interchange navigation.
The same month, Forbes reported that Waymo had imported more than 3,200 Zeekr CM1e vehicles through the Port of Los Angeles since 2024, including more than 2,600 in 2026 alone — a pace of roughly 300 per month. The vehicles arrive with their Chinese connectivity hardware and sensors removed. Waymo installs its own sixth-generation autonomous driving system at a plant in Mesa, Arizona, working with Magna.
Two facts that seem to point in opposite directions. China has 15 million cars with driver-assistance features on the road. America's leading robotaxi company is buying Chinese cars to run its own software. Neither fact says much about who has the better autonomous driving technology. Both describe what each side is actually building.
What 15 Million Actually Measures
The 15 million figure is a stock number, not a flow number. It counts every car on the road that has the hardware and software for NOA, accumulated over years of sales. It does not measure how many cars are sold with the feature this year, and it does not measure how many drivers use it.
The flow numbers tell a different story. According to Zhuoshi Automotive Research, NOA installations in China reached 3.21 million units in the first half of 2026, a penetration rate of 37.4 percent of new passenger car sales. Urban NOA installations in September 2026 reached 498,000 units, a penetration rate of 30.35 percent — the first time the monthly figure crossed 30 percent. The cumulative January-to-September urban NOA penetration rate was 24.9 percent.
Those are strong adoption numbers by any global standard. They are also not the same as 15 million cars driving themselves.
The gap between capability and use is the part the stock figure hides. A survey by Dongwu Securities found that only 31 percent of car owners regularly use urban NOA. 20 percent explicitly said they did not dare to use it. More than 70 percent either rarely used it or never did.
The reasons are straightforward. Urban NOA requires the driver to supervise the system continuously, hands on the wheel or ready to take over. It works well on some roads and poorly on others. It is available in some cities and not in others. And the liability framework for a system that assists but does not replace the driver is still being defined.
A car with NOA hardware and a driver who does not use NOA is a car with a feature, not a car that drives itself. The 15 million figure counts the first. The 31 percent usage rate describes the second.
What China Is Actually Building
The more consequential story is the infrastructure China has built around the feature.
On July 30, 2026, China's State Administration for Market Regulation and Standardization Administration approved GB 44721-2026, a mandatory national standard for the safety requirements of autonomous driving systems. It applies to L3 and L4 vehicles — conditional and high automation — and takes effect on July 1, 2027. A companion standard, GB 47955-2026, covers the safety requirements for combined driving assistance systems and takes effect on January 1, 2027.
These are mandatory standards, not guidance documents. A vehicle that does not meet them cannot be sold.
Alongside the safety standards, China is building an identification regime. On October 20, 2026, the Ministry of Industry and Information Technology published a draft revision to the national standard for vehicle lighting that would create an "ADS marker light" — a blue-green light visible on the exterior of the vehicle, permitted only on L3 and higher vehicles, with implementation targeted for January 1, 2028. The light's function is to tell everyone outside the car who is driving. A blue-green light means the system is in control. No light means the human is.
The regulatory logic is consistent across these instruments. China is not trying to prove that its autonomous driving technology is the best. It is building the legal and physical infrastructure that makes autonomous driving identifiable, auditable, and assignable — a system where a police officer, an insurer, or a court can determine from the outside whether the car or the person was driving at the moment of an incident.
That infrastructure extends to the communication layer. According to MIIT, China has developed more than 400 standards covering vehicle networking, across seven national technical committees. As of August 2026, approximately 16 million passenger cars were equipped with 5G terminals and 2 million with C-V2X terminals — the vehicle-to-everything communication systems that let cars talk to each other and to roadside infrastructure.
The L3 admission pilot is the next step. In December 2025, Changan Automobile and BAIC BluePark received China's first conditional L3 product admission approvals, with test operations in designated areas of Chongqing and Beijing. The second batch of pilot admissions is underway. The Ministry of Industry and Information Technology has said L3 models are expected to open to individual buyers gradually starting in the second quarter of 2026.
The combination is unusual. Most countries have either permissive rules with little enforcement or restrictive rules with little deployment. China has built a mandatory standard, an identification regime, a communication standard, and an admission pilot — while 15 million cars with driver-assistance hardware are already on the road.
Why Waymo Is Buying Chinese Cars
The Waymo purchases are not a vote of confidence in Chinese autonomous driving software. They are a vote of confidence in Chinese vehicle manufacturing economics.
The tariff math explains the decision. The United States imposes a 127.5 percent tariff on Chinese electric vehicles — 2.5 percent base, 100 percent under Section 301, and 25 percent under strategic goods provisions. According to US import declaration data, each Ojai vehicle is valued at about $38,000 at import. With the tariff, the cost rises to about $86,500.
Waymo bought them anyway. The reason is that $86,500 is still roughly half the cost of the Jaguar I-Pace vehicles Waymo had been using, according to industry analysis. The Zeekr platform is purpose-built for autonomous driving — a flat floor, a spacious interior, and a body designed around a sensor suite rather than retrofitted for one. Even after the tariff, the total cost of a Chinese-built robotaxi platform is lower than the alternatives available to Waymo in the United States.
The import structure is designed to satisfy U.S. security rules. The vehicles arrive as bare chassis, with Chinese connectivity hardware and computing systems removed before shipment. Waymo installs its own sixth-generation sensors and compute platform at the Mesa plant. The autonomous driving stack is entirely American. The vehicle platform is Chinese.
That division of labor is the point. Waymo is not buying Chinese autonomy. It is buying a Chinese chassis that happens to be cheaper and better suited to robotaxi conversion than anything American automakers currently offer at scale.
What the Two Facts Mean Together
Put the 15 million figure and the 3,200-vehicle import figure side by side, and the picture that emerges is a division of labor within a single global supply chain rather than a competition between two autonomous driving ecosystems.
China is doing two things. It is deploying driver-assistance hardware at a scale no other market has matched — 61 million L2 vehicles, 15 million with NOA, 16 million with 5G terminals, 2 million with C-V2X. And it is building the regulatory infrastructure that turns those features into a governed system: mandatory safety standards, an exterior marker light that identifies which entity is driving, an L3 admission process, and a standards body that has produced more than 400 vehicle networking specifications.
The United States, through Waymo, is doing something different. It is building the autonomous driving software — the perception, planning, and control stack that actually drives the car — and it is buying the vehicle platform from the country that makes it cheapest.
Neither side is doing what the other is doing. China is not producing a Waymo competitor at scale. Waymo is not producing its own vehicle platform. The 15 million Chinese cars with NOA are running Chinese-developed driver-assistance systems, mostly from automakers and their suppliers. The 3,200 Waymos are running American-developed autonomy on Chinese-built chassis.
That is a supply chain, not a rivalry.
The Part That Is Hard to Replicate
The reason this division of labor is stable — for now — is that each side has built something the other cannot easily copy.
China's advantage in deployment is not a single technology. It is the combination of manufacturing scale, cost structure, and regulatory infrastructure. A Chinese automaker can put NOA hardware into a mass-market vehicle because the sensor and compute costs have fallen far enough, and because the regulatory framework exists to permit the feature. The mandatory standards that take effect in 2027 do not slow deployment; they formalize it. The ADS marker light does not restrict autonomy; it makes autonomy legible.
Waymo's advantage is the software stack and the operational experience. It has driven more autonomous miles than any other company, in more cities, with more safety data. That data is not something a Chinese automaker can replicate by shipping more cars with driver-assistance features. NOA is not autonomous driving. It is a supervised assistance system. The gap between supervised assistance at 15 million units and unsupervised autonomy at commercial scale is the gap Waymo has spent a decade crossing.
The two advantages are complementary. China's deployment scale creates the manufacturing base and the regulatory framework. Waymo's software creates the autonomy. The chassis is Chinese. The brain is American.
Whether that arrangement lasts depends on whether either side decides it needs the other's capability. If a Chinese company builds a competitive autonomous driving stack, Waymo's software advantage narrows. If an American automaker builds a competitive robotaxi platform, Waymo's need for Chinese chassis declines. Neither has happened yet.
What the Numbers Actually Show
The 15 million figure is real, and it is large. It describes a market where driver-assistance hardware has moved from luxury vehicles to the mass market, and where the regulatory system has been built to govern what happens next.
The 3,200 figure is also real, and it is small. It describes one company buying one model of Chinese vehicle because the economics work, even with a 127.5 percent tariff.
The two numbers describe the same industry from different angles. China is building the road, the rules, and the cars. Waymo is building the driver. For now, both are necessary. The question is which one becomes optional first.
Sources: Ministry of Industry and Information Technology at the World Intelligent Connected Vehicles Conference via Securities Times (October 24, 2026), Phoenix Tech (October 24, 2026), and Guandian (October 24, 2026); China Academy of Information and Communications Technology supporting data via CCTV (October 24, 2026); Forbes via TechWeb (August 16, 2026) and 7its (August 17, 2026); Zhuoshi Automotive Research via EET-China (November 1, 2026); Dongwu Securities survey via Gasgoo (July 24, 2026); China Automotive News on urban NOA penetration (October 19, 2026); D1EV on September 2026 urban NOA installations (October 26, 2026); MIIT on GB 44721-2026 and GB 47955-2026 via Guangming Daily (August 4, 2026) and China Standards Online Service (2026); China Industry News on the ADS marker light draft standard (November 1, 2026); MIIT on 400-plus vehicle networking standards via China Industry News (October 22, 2026) and CCTV (October 22, 2026); MIIT on L3 admission pilot via Workercn (October 23, 2026) and Xinhua (October 29, 2026); Sina Finance on Waymo tariff math (August 17, 2026); 12365auto on Zeekr chassis stripping and Mesa assembly (August 17 and 24, 2026); Yahoo Finance on Waymo's stripped-vehicle import structure (August 3, 2026).
Disclaimer
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations
The 15 million figure is a stock count of vehicles sold with the hardware, not a measure of active use.
Usage rates come from a single brokerage survey and may not represent all Chinese drivers.
Waymo import volumes are drawn from reporting based on port data and company statements, not from a public filing.
The U.S. tariff structure and import valuation described here reflect the position as of late 2026.
Reporting standards differ between Chinese and Western outlets, and figures cited here reflect the primary sources as published.
Market size and shipment estimates vary by methodology; where estimates conflict, the source used is named in the text.
The sectors discussed are changing quickly, and some figures may be revised after publication.
Sources
- Ministry of Industry and Information Technology at the World Intelligent Connected Vehicles Conference via Securities Times (October 24, 2026), Phoenix Tech (October 24, 2026), and Guandian (October 24, 2026)
- China Academy of Information and Communications Technology supporting data via CCTV (October 24, 2026)
- Forbes via TechWeb (August 16, 2026) and 7its (August 17, 2026)
- Zhuoshi Automotive Research via EET-China (November 1, 2026)
- Dongwu Securities survey via Gasgoo (July 24, 2026)
- China Automotive News on urban NOA penetration (October 19, 2026)
- D1EV on September 2026 urban NOA installations (October 26, 2026)
- MIIT on GB 44721-2026 and GB 47955-2026 via Guangming Daily (August 4, 2026) and China Standards Online Service (2026)
- China Industry News on the ADS marker light draft standard (November 1, 2026)
- MIIT on 400-plus vehicle networking standards via China Industry News (October 22, 2026) and CCTV (October 22, 2026)
- MIIT on L3 admission pilot via Workercn (October 23, 2026) and Xinhua (October 29, 2026)
- Sina Finance on Waymo tariff math (August 17, 2026)
- 12365auto on Zeekr chassis stripping and Mesa assembly (August 17 and 24, 2026)
- Yahoo Finance on Waymo's stripped-vehicle import structure (August 3, 2026).
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations: The 15 million figure is a stock count of vehicles sold with the hardware, not a measure of active use.; Usage rates come from a single brokerage survey and may not represent all Chinese drivers.; Waymo import volumes are drawn from reporting based on port data and company statements, not from a public filing.; The U.S. tariff structure and import valuation described here reflect the position as of late 2026.; Reporting standards differ between Chinese and Western outlets, and figures cited here reflect the primary sources as published.; Market size and shipment estimates vary by methodology; where estimates conflict, the source used is named in the text.; The sectors discussed are changing quickly, and some figures may be revised after publication.