Chinese autonomous driving company Pony.ai said on Feb. 6 that it has entered into a strategic partnership with Moore Threads, aiming to advance the large-scale deployment of Level 4 autonomous driving using domestically developed AI computing hardware.
Under the agreement, the two companies will work together on training and optimizing Pony.ai’s core technologies, including its world model and virtual driver system. The collaboration will rely on Moore Threads’ MTT S5000 training-and-inference integrated computing cards and its KUAE intelligent computing cluster to support large-scale training, simulation and in-vehicle model development.
See also: Pony.ai Targets Tripling Robotaxi Fleet After Achieving City-Wide Breakeven in Guangzhou
The partnership marks Pony.ai’s first large-scale adoption of Chinese-developed AI computing for critical autonomous driving workloads, while also representing Moore Threads’ entry into the core computing infrastructure used for autonomous vehicle development. Both companies said the collaboration is intended to create closer integration across algorithms, data, computing resources and applications, with the goal of shortening development cycles and reducing costs.
Pony.ai has developed its autonomous driving system around a self-built world model, known as PonyWorld, and a virtual driver framework that uses reinforcement learning to generate and train on large volumes of simulated driving data. The company said the system can produce billions of kilometers of simulated test scenarios each week, including variations of high-risk situations, to support continuous model iteration.
Moore Threads, one of a small number of Chinese companies developing full-featured GPUs, has focused on AI computing, scientific workloads and graphics rendering. Its KUAE computing cluster is designed to support large-scale model training, while its graphics capabilities are used for high-fidelity simulation and scenario reconstruction, which are key components of autonomous driving development.
See also: SANY and Pony.ai Autonomous Heavy-Duty Truck Nears Mass Production
The partnership aligns with Pony.ai’s broader commercialization roadmap. By the end of 2025, the company said its robotaxi fleet had exceeded 1,100 vehicles, surpassing its annual deployment target. Its seventh-generation robotaxi system has also reached positive unit economics in Guangzhou, according to the company.
Pony.ai recently announced that the first mass-produced bZ4X robotaxi developed with Toyota has rolled off the production line, marking another step toward scaling commercial robotaxi operations. The company has also expanded partnerships with automakers and mobility service providers, including BAIC BJEV and ATBB, as it works toward broader deployment of Level 4 autonomous driving services.