PHYMI Raises Nearly $100M in Seed Funding to Build Embodied AI Agents

PHYMI, an embodied AI startup founded by former DeepRoute.ai executive Liu Nianqiu, has raised nearly US$100 million in seed funding. The deal ranks among the largest seed rounds for a Chinese embodied AI company this year, signaling strong investor appetite for robotics and physical intelligence.

Liu’s background in autonomous driving gives PHYMI a leadership team familiar with real-world perception, planning, and safety-critical systems. DeepRoute.ai is known for developing autonomous driving technology, and that experience may inform how PHYMI approaches embodied systems.

The company has not yet detailed its first commercial product or launch schedule. However, the size of the seed round suggests investors are betting on a long-term platform opportunity rather than a near-term application.

Embodied AI has become a hot category for investors in 2025, as advances in large language models and multimodal learning extend to physical systems. PHYMI’s seed round reflects that momentum, though the company remains early in its development.

The round comes amid a wave of funding for robotics and AI startups in China, even as broader venture markets remain selective.

IDG Capital led the round, with participation from Yunqi Capital, DiDi, Fosun RZ Capital, Glory Ventures, and Hesai Technology. The diverse investor group includes venture capital firms, a ride-hailing giant, and a lidar maker, suggesting interest across the autonomy and robotics supply chain.

PHYMI is developing 'Physical Agents' designed to understand goals, move and manipulate objects, and complete tasks in open, dynamic environments. Unlike narrow automation systems that repeat predetermined motions, these agents are intended to adapt to unstructured settings, a long-standing challenge in robotics.

Founded earlier this year, the company aims to combine advances in foundation models with embodied control. That approach reflects a broader industry shift toward general-purpose robots that can learn from data rather than rely on hand-coded rules.

The involvement of DiDi, a major mobility platform, and Hesai, a lidar manufacturer, could give PHYMI strategic access to transportation data and sensor hardware. Such partnerships may help the startup validate its technology in real-world conditions.

The term 'Physical Agents' is not yet standardized, but it generally refers to AI-driven entities that can operate in the physical world with some degree of autonomy. PHYMI’s description suggests a focus on goal-directed behavior, spatial reasoning, and dexterous manipulation.