Skip to content

Embodied Intelligence Company Landscape

Overview

Embodied intelligence is at a critical stage of transitioning from laboratories to industrialization. Globally, dozens of companies are developing humanoid robots, quadruped robots, dexterous manipulation systems, and foundation model platforms. This article surveys major companies worldwide and in China, covering their technical approaches, funding status, and product progress.

Major Global Companies

Company Overview Table

Company Founded HQ Key Product Core Technology Funding Scale Current Status
Tesla 2003 Texas, US Optimus End-to-end neural networks, FSD transfer, Dojo supercomputer Self-funded Internal factory testing
Figure 2022 California, US Figure 02 VLA large models, OpenAI collaboration, whole-body control ~$675M (val. $2.6B) BMW production line pilot
1X Technologies 2014 Norway NEO Beta RL, whole-body compliant control ~$125M (OpenAI led) NEO Beta prototype demo
Agility Robotics 2015 Oregon, US Digit Bipedal walking, logistics handling ~$179M Amazon warehouse pilot
Apptronik 2016 Texas, US Apollo NASA background, modular design ~$75M Mercedes-Benz collaboration
Sanctuary AI 2018 Vancouver, Canada Phoenix Carbon (AI control system), dexterous hands ~$150M Commercial pilot
Boston Dynamics 1992 Massachusetts, US Atlas (electric) Hydraulic/electric drive, whole-body dynamics Hyundai acquisition ($1.1B) Electric Atlas in development
Physical Intelligence 2024 San Francisco, US pi0, pi0.5 General robot foundation model, VLA ~$400M (val. $2.4B) Model release and partnerships

Key Company Details

Tesla Optimus

Core advantages: - Data flywheel: FSD visual understanding transferable to robotics - Vertical integration: Self-developed chips (Dojo/HW5), motors, batteries, inference hardware - Manufacturing at scale: Automotive factory mass production experience - Cost target: Elon Musk claims ultimate price below $20,000

Physical Intelligence

Founded in 2024 with massive funding, founding team from Google DeepMind and UC Berkeley: - pi0: First general robot foundation model, VLA architecture - pi0.5: Improved version with significantly better cross-task generalization - Core team: Karol Hausman, Sergey Levine, and other renowned researchers

Major Chinese Companies

Company Overview Table

Company Founded HQ Key Product Core Technology Funding Current Status
Unitree 2016 Hangzhou H1/G1 (humanoid), Go2 (quadruped) High-performance motors, low-cost quadrupeds Hundreds of millions RMB G1 mass production, H1 iteration
Fourier Intelligence 2015 Shanghai GR-1/GR-2 Rehabilitation robot background, force control Hundreds of millions RMB GR-2 released, mass production prep
UBTECH 2012 Shenzhen Walker S Full-stack humanoid robot, HKEX listed HKEX IPO (2023) Walker S factory pilot
Agibot 2023 Shanghai Expedition A2 Huawei background, end-to-end AI ~$100M+ Rapid iteration
Galbot 2023 Beijing Galbot G1 Dexterous manipulation, general grasping Hundreds of millions RMB Prototype demo
XPeng Robotics 2020 Guangzhou PX5 XPeng Auto technology synergy XPeng ecosystem investment PX5 released
Noetix 2023 Beijing STAR1 Tsinghua-origin, high-dynamic motion Hundreds of millions RMB Prototype demo
Engineer AI 2023 Shanghai Dexterous manipulation system Foundation model + manipulation Hundreds of millions RMB R&D
Deep Robotics 2017 Hangzhou Jueying X30 Quadruped robots, industrial inspection Hundreds of millions RMB Commercial deployment
Agile Robots 2018 Beijing/Munich Diana series DLR background, force-controlled collaboration ~$220M Commercial deployment

Company Ecosystem Map

graph TB
    subgraph Humanoid["Humanoid Robots"]
        Tesla[Tesla Optimus]
        Figure[Figure 02]
        OneX[1X NEO]
        Agility[Agility Digit]
        Unitree_H[Unitree H1/G1]
        Fourier[Fourier GR-2]
        UBTECH[UBTECH Walker S]
        Agibot[Agibot Expedition A2]
    end

    subgraph Quadruped["Quadruped Robots"]
        Unitree_Q[Unitree Go2/B2]
        DeepRobotics[Deep Robotics Jueying]
        BD_Spot[Boston Dynamics Spot]
    end

    subgraph Dexterous["Dexterous Manipulation"]
        Galbot[Galbot]
        Engineer[Engineer AI]
        AgileRobots[Agile Robots]
    end

    subgraph Foundation["Foundation Models"]
        PI[Physical Intelligence pi0]
        Google_RT[Google RT-2/RT-X]
        NVIDIA_GR[NVIDIA GR00T]
    end

    PI --> Figure
    PI --> Unitree_H
    NVIDIA_GR --> Agility

Competitive Landscape Analysis

Technology Route Comparison

Dimension US Companies Chinese Companies European Companies
AI Capability Leading (large model ecosystem) Rapidly catching up Deep academic accumulation
Hardware Cost Higher Significant advantage Medium
Manufacturing Tesla leads Complete supply chain Precision manufacturing
Commercial Deployment Factory pilots mainly Multi-scenario exploration Industrial applications
Funding Environment Large-round funding active Policy + capital dual-driven Relatively conservative
  1. Foundation model empowerment: Nearly all companies integrating VLA and other large model technologies
  2. Cost competition: Chinese companies have significant hardware cost advantages, driving global price decline
  3. Scenario focus: Shifting from general-purpose to specific scenarios (factories, logistics, inspection)
  4. Ecosystem competition: NVIDIA (Isaac/GR00T), Google (RT-X) and other platform ecosystem battles
  5. Policy-driven: China has listed humanoid robots as a strategic emerging industry with nationwide support policies

评论 #