Alibaba Teams up with Nvidia on “Physical AI,” Integrating Full-Stack Tools into Alibaba Cloud’s PAI Platform
Want to read in a language you're more familiar with?
Alibaba teams with Nvidia on “Physical AI,” wiring Nvidia’s stack into PAI—and pledges a $53B, 3-year AI infra buildout to speed embodied-AI apps.
Alibaba announced a wide-ranging Physical AI collaboration with Nvidia at the 2025 Apsara (Yunqi) Conference on September 24 2025, aiming to speed up real-world, embodied-intelligence applications from robotics to assisted driving. The partnership spans synthetic data generation, model training, simulation-based reinforcement learning, and model validation and testing, the company said.
Alibaba Cloud’s PAI (Platform for Artificial Intelligence) will integrate Nvidia’s Physical AI software stack to offer enterprise customers an end-to-end toolchain covering data preprocessing, simulation data creation, training and evaluation, robot RL, and simulation testing—designed to shorten development cycles for embodied agents and ADAS-style systems.
Framing the strategic context, Alibaba Group CEO Wu Yongming said large models are “the next operating system,” with “AI cloud” becoming “the next computer,” arguing that only five or six super cloud platforms may ultimately dominate. He added that Alibaba is executing a RMB 380 billion ((about $53 billion) three-year AI infrastructure plan, with potential for further investment. By 2032, the company expects the power consumption of Alibaba Cloud’s global data centers to be 10× 2022 levels to meet AI demand. Alibaba’s Hong Kong-listed shares jumped more than 9% after the remarks, local media noted.
“Physical AI”—systems that perceive, understand spatial relationships and physics, and act in the real world—has become a focal point as cloud providers and chipmakers race to move beyond pure text/image generation. By wiring Nvidia’s stack into PAI, Alibaba is positioning itself as a one-stop environment for synthetic-to-real development: simulate, train, validate, and deploy in robotics and autonomous systems. Alibaba did not disclose commercial timelines or specific customer pilots, but the companies say the integrated toolchain is intended to reduce time-to-market for embodied-AI applications across industrial automation, logistics robotics, and driver-assistance use cases.
