Robotaxis Are Not About Technology — They’re About Operations

Waymo continues to set the benchmark for “the safest autonomous driving,” expanding city by city with a strong focus on technology and validation. Cruise pursued rapid scaling, but the large-scale pullback following its 2023 accident exposed the limits of that strategy. Pony.ai, by contrast, is redefining robotaxis as an industrial system, placing cost structure, operational capability, and full lifecycle management at the center of commercialization.
At Automechanika Shanghai, Pony.ai Vice President Wang Qiang emphasized that the future of autonomous driving rests not on technological novelty alone, but on practical realities—operations, cost, and end-to-end system design. Not the performance of a single vehicle, but the ability to operate thousands simultaneously. Not a single trip, but a lifecycle spanning vehicle introduction, operation, and retirement. This is where Pony.ai is outlining what it takes for autonomous driving to become a true industry.

By Sang Min Han _ han@autoelectronics.co.kr
한글로보기







“We spoke directly with robotaxi operators and heard many practical questions and voices from the field. These insights represent an essential step in discussing the commercialization of autonomous driving going forward.”
— Wang Qiang, Vice President, Pony.ai

When autonomous driving tests are discussed, simulations and closed test tracks often come to mind. Robotaxis, however, can be seen as another form of test track—one conducted on real roads. Because they operate in live traffic environments, the demands for safety and reliability are far more complex and multi-layered. This is why many companies are expanding beyond the US and China into regions such as the Middle East and Europe, broadening the scope of validation and real-world deployment.

But what does commercialization of robotaxis really mean? Wang offers a clear answer.


 

“Commercialization is not achieved simply by launching an impressive vehicle or declaring safety compliance. It is far more complex. Technical maturity must align with policy, operational capability, and cost structure.”

China is the world’s largest autonomous driving test environment, allowing robotaxi testing across more than 20,000 km² of urban areas and over 20,000 km of roads. Pony.ai has conducted robotaxi trials in limited regions, with some vehicles already approved for real services on public roads. Testing has taken place in cities including Beijing, Shanghai, Guangzhou, and Shenzhen, resulting in the accumulation of more than 55 million kilometers of global driving data.

Pony.ai continuously trains its systems through simulation and data-driven learning to closely replicate natural human driving behavior. This is not merely about passenger comfort—it is essential for safety and trust. In certain situations, the system must intervene autonomously and initiate a takeover, always following the principle of achieving a minimal risk state. These interventions, in turn, generate critical data for further system improvement.

“Once the technical foundation is in place, policy guidance and regulatory oversight become essential. The Chinese government has designated specific regions as robotaxi pilot zones, and revisions to road traffic laws and autonomous driving regulations have further underscored the importance of AI-based driving technologies. Autonomous driving is included in China’s 15th Five-Year Plan, with AI development clearly identified as a core priority.”






 
When commercialization is considered seriously, cost becomes a decisive factor.

“Why is cooperation with OEMs so important? Because cost determines commercialization. Without cost competitiveness, commercialization cannot happen. Even when technology and policy are ready, the final barrier is operational capability.”

By leveraging core platforms provided by partner OEMs, Pony.ai has reduced overall system costs by as much as 70%, with further reductions remaining a key objective. As operational scale increases tenfold or twentyfold, cost structure becomes even more critical.

One of Pony.ai’s most important partners is Toyota. The two companies began technical collaboration in 2019, producing their first joint test vehicle and securing major investment the following year. In 2023, Pony.ai established GTMC (GAC Toyota), a joint venture for mass production of Level 4 robotaxis, integrating Pony.ai’s autonomous driving software into Toyota’s electric vehicle platform. These Gen-7 robotaxis are now being deployed at scale.

“Passengers have diverse expectations. Vehicles incorporate voice authentication for identity verification, ambient lighting systems, and sensors throughout the cabin that predict collision risks and send alerts. Operations are not about making a single vehicle move well. At the commercialization stage, the core challenge is managing tens of thousands of vehicles simultaneously.”

Vehicles that autonomously leave depots to pick up passengers, navigate to charging stations when needed, and return to base after service—these capabilities define the capacity to scale. Commercialization ultimately means managing an entire system in which thousands of vehicles operate autonomously, not just proving the performance of one.


 
Under such an operational model, conventional vehicle design is no longer sufficient. Traditional vehicles remain centered on human drivers—brake pedals, accelerator pedals, and legacy powertrain layouts. In the robotaxi era, vehicles must be designed from the ground up.

This is why Pony.ai emphasizes close collaboration with OEMs starting at the R&D stage, while considering the full vehicle lifecycle—from deployment to operation and eventual retirement.

“For example, when a Level 2 vehicle reaches the end of its life, how should it be dismantled? How should sensors and data storage devices be handled? Everything must be redefined within a lifecycle management framework,”
Wang explained.

Pony.ai’s collaboration model is built on co-growth with OEMs, taxi operators, and platform partners. Defining itself as a technology company, Pony.ai works with Toyota, Guangzhou Automobile Group (GAC), Beijing Automotive Group (BAIC), and others, while also partnering with taxi companies across multiple regions and expanding internationally.

China and the United States remain key markets, but Pony.ai brings its technology wherever demand exists. In China, more than 1,000 seventh-generation robotaxis are already in commercial operation. In Europe, Pony.ai is collaborating with Stellantis to co-develop an electric LCV-based robotaxi, with real-world road testing scheduled to begin in Luxembourg in 2026. Through partnerships with Uber, pilot services are being prepared in regions such as the Middle East, allowing users to hail robotaxis via the Uber app.

Pony.ai has also entered Korea, co-developing smart driving solutions for production vehicles with Horizon Robotics, and integrating its services with platforms such as WeChat, DiDi, and GDO, enabling easier access and user feedback.

“Pony.ai wants to share the value of technology with its partners. We aim to share both the achievements and the benefits of technology, and to build the automotive ecosystem together. That is our message.”

 

---

Automechanika Shanghai 2025 and What Comes After ‘China for China’:
CARIAD China’s Perspective on Autonomous Driving:

[AEM] Automotive Electronics Magazine

---