PHYMI Raises Nearly $100 Million in Seed Funding for Physical Agents

The new startup plans to use the funding for research, real-world data systems, engineering, deployment and hiring—turning its physical-AI ambition into an execution test.

By 2 min read
PHYMI Raises Nearly $100 Million in Seed Funding for Physical Agents
PHYMI Raises Nearly $100 Million in Seed Funding for Physical Agents

Listen to this story

The audio brief

About 1:30
0:001:30
Read transcript
PHYMI has raised nearly 100 million dollars in seed funding to build AI systems that can operate in the physical world, not just inside software. IDG Capital led the round, with participation from Yunqi Capital, Didi, Fosun RZ Capital, Glory Ventures, and Hesai Technology. For a newly founded startup, that is substantial financing—and it sets an unusually demanding execution test. PHYMI calls its systems Physical Agents: embodied AI foundation models designed to understand goals, perceive their surroundings, navigate changing environments, and manipulate objects. The distinction matters. A software agent can complete a task on a screen; a physical agent has to act when conditions are open, dynamic, and not fully predictable. PHYMI says it will use the money for core research, real-world data systems, product engineering, deployment, and hiring. That combination suggests the company is trying to build the full stack, from collecting useful physical-world experience to turning research into products. The founder is Liu Nianqiu, who spent more than seven years at autonomous-driving company DeepRoute.ai. There, according to DealStreetAsia, he worked on map-free autonomy and on the shift toward end-to-end Vision-Language-Action systems. That background is relevant, but the funding announcement is evidence of commitment, not proof that the agents already work. The key question now is whether PHYMI’s spending on data, engineering, and deployment produces working products rather than only ambitious research.

Story brief

3 key points

PHYMI has secured nearly $100 million in seed financing, an unusually large early round led by IDG Capital, to build embodied AI systems that can reason about goals and act in changing physical environments. The startup plans to spend the capital on core research, real-world data infrastructure, product engineering, deployment, and hiring. Founded earlier this year by former DeepRoute.ai executive Liu Nianqiu, PHYMI...

  1. 01

    The round includes Yunqi Capital, Didi, Fosun RZ Capital, Glory Ventures, and Hesai Technology.

  2. 02

    PHYMI’s target is broader than software agents: perception, navigation, and object manipulation in open, dynamic settings.

  3. 03

    Liu spent more than seven years at DeepRoute.ai, working on map-free autonomy and Vision-Language-Action architectures.

PHYMI now has nearly $100 million to pursue AI systems meant to operate beyond the screen: understanding goals, moving through changing settings and manipulating objects. The newly founded startup raised the seed round with IDG Capital as lead investor, giving an early company substantial capital for building physical-world AI.

The investor group includes Yunqi Capital, Didi, Fosun RZ Capital, Glory Ventures and Hesai Technology alongside IDG Capital. DealStreetAsia described Didi as a ride-hailing company and Hesai as a 3D lidar maker.

The company’s bet is on AI with a body

PHYMI describes its systems as Physical Agents. In the company’s framing, they are embodied AI foundation models that pair reasoning with physical actions in open, dynamic environments. That is a different target from an agent confined to software: PHYMI is aiming for systems that can perceive, navigate and manipulate objects as conditions change around them.

Where PHYMI says the money will go

  • Core research for its Physical Agent AI offerings.
  • Dynamic data systems built from real-world settings.
  • Product engineering and deployment work.
  • Hiring to expand the company’s team.

Those priorities put research, data infrastructure and product-building on the same financing plan. Its goal-oriented agents would need both the ability to interpret a task and the ability to carry out physical movement and object manipulation in environments that are not fixed or tightly controlled.

A founder from autonomous driving

PHYMI was founded earlier this year by Liu Nianqiu, a former DeepRoute.ai executive. DealStreetAsia reported that Liu spent more than seven years at the autonomous-driving company as a vice president and partner, where he helped with map-free autonomous-driving deployments and its shift toward end-to-end and Vision-Language-Action architectures.

That background connects PHYMI’s current effort to a field where systems must make decisions amid moving, imperfectly predictable conditions. But the financing announcement is a commitment to build, not evidence that PHYMI’s intended agents can already complete such work. The immediate test is whether its investment in research, data systems, engineering and hiring can turn the stated physical-world ambitions into deployed products.

Sources

  1. technode.comEmbodied AI startup PHYMI raises nearly US$100 million in seed funding · TechNode
  2. dealstreetasia.comIDG Capital leads near-$100m seed round for nascent embodied AI startup PHYMI

Loading discussion...