In a significant leap forward in the commercialization of autonomous driving technology, Pony.ai has officially unveiled its next-generation domain controller for autonomous driving, developed in close collaboration with NVIDIA. The new system, announced on Monday, April 27, 2026, is built upon the advanced NVIDIA DRIVE Hyperion architecture and is powered by the formidable DRIVE AGX Thor platform. This strategic partnership underscores the escalating computing requirements for reliable Level 4 autonomous driving capabilities and highlights the deepening integration between Chinese autonomous vehicle developers and leading American semiconductor manufacturers.
Unprecedented Computing Power
The newly unveiled domain controller represents a massive upgrade in processing capabilities, designed specifically to handle the immense data loads generated by Level 4 autonomous driving systems. By incorporating NVIDIA’s NVLink technology, the system enables high-speed communication between two DRIVE Thor system-on-chips (SoCs), facilitating seamless data transfer and processing. This multi-chip configuration delivers a combined peak computing capacity of an astonishing 4,000 FP4 teraflops (TFLOPS), providing the computational muscle required for complex multi-sensor data fusion, real-time environment perception, and sophisticated scenario interpretation.
James Peng, the Founder and CEO of Pony.ai, emphasized the critical nature of this technological advancement. “Our collaboration with NVIDIA has supported several critical milestones in Pony AI’s autonomous driving journey,” Peng stated. “The next-generation domain controller built on NVIDIA DRIVE Hyperion will be a key enabler for the continued evolution of our L4 autonomous driving products and help accelerate large-scale commercialization.” This statement reflects the industry’s broader consensus that raw computing power is the primary bottleneck preventing the widespread deployment of fully autonomous vehicles.
The partnership between Pony.ai and NVIDIA is not a recent development; the two companies have been collaborating since 2017. This long-standing relationship has yielded significant results, most notably in 2025 when Pony.ai commenced mass production of its L4 robotaxi controller, which was powered by four NVIDIA DRIVE AGX Orin chips. This seventh-generation robotaxi fleet demonstrated the viability of NVIDIA’s hardware in real-world applications, paving the way for the adoption of the more advanced DRIVE Thor platform.
Accelerating Global Deployment
The introduction of the new domain controller is a critical component of Pony.ai’s aggressive expansion strategy. The company has already seen substantial growth, with shipments of its “Fangzai” domain controller rising by more than 500% year over year in 2025. These systems have been successfully deployed across a diverse range of applications, including last-mile delivery, robotic street cleaning, logistics, mining, and autonomous shuttles, in countries such as Germany, the United Kingdom, South Korea, Japan, and Switzerland.
Looking ahead, Pony.ai has set ambitious targets for its robotaxi operations. The company plans to expand its fleet to more than 3,000 vehicles and operate in over 20 cities worldwide by the end of 2026. The deployment of the NVIDIA DRIVE Hyperion-based domain controller will be instrumental in achieving these goals, providing the necessary reliability and performance to ensure safe and efficient autonomous operations at scale. This expansion will not only solidify Pony.ai’s position as a leader in the autonomous vehicle sector but also serve as a critical proving ground for NVIDIA’s latest automotive hardware.
(Related: WeRide and Grab Launch Public Robotaxi Operations in Singapore)
The Competitive Landscape
The unveiling of Pony.ai’s new platform comes amid an increasingly competitive global autonomous vehicle market. As companies race to achieve Level 4 autonomy, the reliance on high-performance computing hardware has become a defining characteristic of the industry. Yahoo Autos reported that integrating NVIDIA’s DRIVE Thor platform gives Pony.ai a significant competitive advantage, enabling it to process complex sensor data more efficiently than many rivals.
This development also highlights the complex dynamics of the US-China technology relationship. While geopolitical tensions continue to impact the semiconductor industry, the collaboration between Pony.ai and NVIDIA demonstrates that cross-border partnerships remain essential for advancing cutting-edge technologies like autonomous driving. As Chinese companies continue to push the boundaries of AI applications, their reliance on advanced American hardware underscores the interconnected nature of the global technology ecosystem. The success of Pony.ai’s ambitious expansion plans will likely depend on its ability to navigate these geopolitical complexities while continuing to leverage the best available technology.
The Path to Commercialization
The commercialization of Level 4 autonomous driving is a multifaceted challenge that extends beyond raw computing power. It requires a robust ecosystem of sensors, software algorithms, and regulatory frameworks. Pony.ai’s integration of the NVIDIA DRIVE Hyperion architecture is a crucial step in addressing the hardware component of this equation. The platform’s ability to support multi-sensor data fusion is particularly important, as it allows the vehicle to process inputs from LiDAR, radar, and cameras simultaneously, creating a comprehensive and highly accurate representation of the surrounding environment.
This enhanced perception capability is essential for navigating complex urban scenarios, where unpredictable elements such as pedestrians, cyclists, and other vehicles require split-second decision-making. The 4,000 FP4 TFLOPS of computing capacity provided by the dual DRIVE Thor SoCs enables the autonomous driving system to process this massive influx of data in real time, minimizing latency and maximizing safety. As Pony.ai prepares to scale its robotaxi fleet to 3,000 vehicles by the end of 2026, the reliability and performance of this underlying hardware will be paramount.
Furthermore, the deployment of this advanced domain controller is expected to significantly impact the overall cost structure of autonomous vehicle operations. By consolidating multiple processing functions onto a single, highly efficient platform, Pony.ai can reduce the complexity and power consumption of its onboard systems. This optimization is critical to achieving the economic viability required for large-scale commercialization, as it directly affects the total cost of ownership for fleet operators. As the industry moves closer to widespread adoption, the ability to deliver high-performance autonomous driving capabilities at a competitive price point will be a key differentiator for companies like Pony.ai.