Original Author: snow
Abstract
This report explores how technology trends dominated by traditional tech giants are migrating to the cryptocurrency market and predicts that this cyclical pattern will reappear in the humanoid robotics sector. The report first reviews the evolutionary paths of the metaverse and artificial intelligence waves, then analyzes the investment dynamics of global tech capital in embodied intelligence, and introduces the progress and challenges of top humanoid robotics companies. Finally, it focuses on an overview of Web3 robots and embodied intelligence projects, covering key project features, economic models, and development potential. Through in-depth analysis, this report aims to provide industry participants with cutting-edge insights, revealing the strategic significance of a decentralized machine economy in connecting the physical and digital worlds.
1. The cyclical pattern of traditional tech giants defining trends, Wall Street capital quickly following, and the Crypto market efficiently replicating and evolving into decentralized investment opportunities may reappear in the humanoid robotics sector.
1. Review of the Migration from Tech Giants to the Mainstream Narrative of Crypto
The evolution of the Web3 narrative is heavily influenced by the movements of traditional tech giants and financial capital. Looking back at the past few narrative waves, the core path can be summarized as: traditional tech giants define trends, Wall Street capital quickly follows, and the Crypto market efficiently replicates and evolves into decentralized investment opportunities.
Metaverse Wave (2021–2022):
Meta (formerly Facebook) was the first to rename itself in October 2021 and announced a comprehensive transformation into the metaverse, investing tens of billions of dollars to develop the VR/AR ecosystem; Microsoft launched the Mesh for Teams mixed reality platform in November of the same year, focusing on enterprise-level virtual collaboration; Nvidia released the Omniverse digital twin platform around the same time; Qualcomm established a $100 million fund and launched Snapdragon Spaces to expand the AR ecosystem. This series of actions marked the comprehensive layout of internet giants in the virtual world.
The performance of crypto assets was also significant, with tokens from metaverse platforms like Decentraland (MANA) and The Sandbox (SAND) increasing by 70 times and 280 times, respectively. The floor prices of virtual land NFTs such as Decentraland LAND and The Sandbox LAND once soared to around 5 ETH and 4 ETH, equivalent to tens of thousands of dollars. Leading NFT projects like Bored Ape Yacht Club (BAYC) surged from an issuance price of 0.08 ETH to 150 ETH, valued at over $400,000. The metaverse assets during this period achieved unprecedented short-term explosive growth driven by celebrity effects, major companies entering the market, and speculative fervor, although the market subsequently underwent a significant correction, the astonishing gains still became a landmark event in crypto history.
AI Wave (2023–2024):
At the end of 2022, OpenAI launched ChatGPT, igniting a global AI craze and marking the arrival of the generative artificial intelligence era. Microsoft quickly responded by announcing an additional $10 billion investment in OpenAI in January 2023 and deeply integrating its technology into the Bing search engine and the Office suite. Google followed suit by launching Bard conversational AI (later upgraded to Gemini) and reorganizing the DeepMind team to accelerate AI research and development. Meta shifted to an open-source strategy, releasing the Llama series of large language models to promote industry ecosystem development.
Starting in 2023, several AI narrative tokens gained funding support, with platform tokens like Fetch.ai (FET) and Render Token (RENDER) showing significant price increases, rising over 10 times by early 2024 to set historical highs. Decentralized AI projects like Bittensor (TAO) attempted to build decentralized neural network model training and incentive mechanisms, with their token prices outperforming mainstream assets, yielding over tenfold returns in the following period. By the end of 2024, the market was driven by the trends of AI Agents and related infrastructure, giving rise to several hundred-fold opportunities in a short time, completely igniting the crypto market's FOMO sentiment towards the AI sector.
2. Global Tech Capital Increases Investment in Embodied Intelligence: Humanoid Robots Become the Next Narrative Focus
Currently, the attention of traditional tech giants and top venture capitalists towards humanoid robots is rapidly increasing, and this "technology trend—capital influx—narrative migration" chain reaction model is being replayed, shifting the focus from AI and the metaverse to embodied intelligence and humanoid robots, becoming a core candidate for the next round of Web3 narratives.
Morgan Stanley: Humanoid Robots Will Reshape the Global Labor Market
In February 2025, Morgan Stanley released a research report titled "Humanoid 100: Mapping the Humanoid Robot Value Chain," predicting that approximately 75% of jobs in the U.S. (63 million positions) are "humanoid robot adaptable," potentially affecting wage expenditures by up to $2.96 trillion by 2050. Globally, the potential market size for this industry could reach $9 trillion.
The report divides the industry chain into three core modules:
Brain: Covers foundational AI models (such as NVIDIA's Project Groot), data simulation and modeling (such as Palantir, Oracle), visual systems, and semiconductors (NVIDIA, Intel, Qualcomm, etc.).
Body: Composed of actuators (NSK, RBC Bearings), sensors (Analog Devices, Robosense), and battery systems (CATL, LG Energy), focusing on lightweight design.
Integrators: Includes companies like Tesla, Apple, Samsung, Xiaomi, Amazon, Alibaba, etc., capable of building complete robotic systems.
Goldman Sachs: 2025 May Become the Year of Mass Production for Humanoid Robots
In its February 2025 report "Humanoid Robots III," Goldman Sachs pointed out that humanoid robots are entering a "supply chain leap period," with Asian manufacturing companies rapidly entering the field, and the global industrial ecosystem beginning to take shape. It is expected that by 2035, the annual scale of the industry could reach $38 billion, and in an optimistic scenario, it could exceed $200 billion.
Among them, companies like 1x (Norway), Figure AI (USA), Agility Robotics (USA), Yushu Technology (China), AGIBOT (China), and Leju Robotics (China) have already launched commercial products; Tesla (USA), Sancturay AI (Canada), Boston Dynamics (South Korea), Apptronik (USA), UBTECH (China), Kepler (China), Xiaopeng (China), Fourier (China), Galaxy General (China), and Zhejiang Humanoid Robot Innovation Center (China) are in the factory trial production stage.
Tesla: Robots Will Become a "Trillion-Dollar" Growth Engine
On January 29, 2025, during Tesla's Q4 2024 earnings call, Tesla CEO Elon Musk stated that humanoid robots will become the main force in industry, with numbers expected to surpass humans, potentially reaching 10 billion to 20 billion units, with a goal of producing 1 billion units per year, capturing over 10% of the market, which would bring Tesla a market value of $25 trillion to $30 trillion.
Nvidia: The ChatGPT Moment for General Robots is Coming Soon
At COMPUTEX 2025 held on May 21, 2025, Nvidia predicted that within 3 to 5 years, specialized robots will be prioritized for widespread adoption (in manufacturing, logistics, etc.), but limited by non-technical factors (safety certification, social acceptance); in about 10 years, general robots may become part of daily life, needing to overcome technical bottlenecks such as hardware clustering, simulation efficiency, and data integration, relying on "economies of scale."
Nvidia CEO Jensen Huang announced the launch of Isaac GR00T N1 at the 2025 GTC conference, which is the first open-source robotic foundational model platform, adopting a dual mechanism architecture that mimics human reflexes and reasoning. At the same time, the Cosmos simulation platform and Newton physics engine were released to generate high-quality training data, accelerating the interaction capabilities of robots with the real world.
Additionally, Nvidia's launch of the Blue entertainment robot indicates that robots are no longer limited to industrial scenarios and are entering the consumer and interactive experience fields.
3. Overview of Global Top Humanoid Robotics Companies
(1) Tesla Optimus
Tesla's general humanoid robot Optimus focuses on factory tasks and home applications, relying on pure visual AI and FSD technology, with plans for mass production in 2025 at a target price of under $30,000.
Development History
August 2021: The first AI Day released the Optimus concept image, announcing the humanoid robot plan for the first time.
October 2022: The initial bare-bones version was showcased at AI Day.
March-December 2023: The robot achieved walking, picking up objects, visual recognition, simple actions (such as yoga), and end-to-end neural network training.
February-October 2024: The Gen 2 version was released, enhancing sensors and controllers, with more stable gait and capabilities for autonomous navigation, charging, climbing stairs, and interacting with people.
Starting in 2025: Plans to test thousands of units internally; mass production of 50,000 to 100,000 units by 2026, reaching 500,000 to 1 million units by 2027.
Technical Features
Structural Performance: Gen 2 is approximately 1.73 meters tall, weighs 57 kg, and can carry 20 kg; Gen 3 is equipped with a 22-degree-of-freedom dexterous hand.
Power System: Uses electric vehicle batteries and drive systems, providing long endurance and high efficiency.
Motion Capabilities: Supports basic functions such as walking, squatting, picking up objects, and standing on one leg.
Perception and Control: Utilizes the FSD computing platform, supporting autonomous navigation and environmental perception.
Execution Structure: 14 rotary joints (frameless motors + harmonic reducers), 14 linear joints (planetary roller screw), totaling 70 bearings.
Main Achievements
The Gen 2 prototype has achieved basic industrial actions and autonomous navigation.
The FSD AI platform has been successfully ported to the robotic system.
Internal Pilot: 1,000 units to be deployed in Tesla factories starting in 2024.
Market Planning: Thousands of units to be produced in 2025, with an expected price of $20,000 to $30,000.
Long-term Goal: Achieve a scale of 1 billion units, replacing repetitive human labor.
(2) Figure AI
Founded by Brett Adcock in 2022, this American robotics company focuses on developing AI-driven humanoid robots aimed at addressing labor shortages, particularly in manufacturing and retail.
Development History
May 2022: The company was founded, focusing on the development of humanoid robots.
October 2023: Released Figure 01, integrating GPT-4 to achieve natural language interaction and object classification.
February 2024: Completed Series B financing of $675 million (investors include Microsoft, Nvidia, OpenAI, etc.), with a valuation of $2.6 billion.
August 2024: Released Figure 02, with comprehensive hardware upgrades, featuring a 16-degree-of-freedom dexterous hand and a 20-hour battery life.
November 2024: 40 units of production demand, with an expected output of 2,000 units by April 2025 and a target of 20,000 units in Q3.
February 2025: Figure AI is seeking $1.5 billion in Series C financing, with a valuation of approximately $40 billion.
Future plans: Aim to ship 100,000 units within four years, with a new production base set to be operational in 2025.
Technical Features
Vision and Perception: Equipped with 6 RGB cameras and LiDAR, providing comprehensive environmental perception capabilities.
Hand Capabilities: 16 degrees of freedom in a single hand, capable of bearing 25 kg, with high precision.
Computing Platform: Equipped with NVIDIA RTX GPU, with inference capabilities three times that of the previous generation.
AI System: Utilizes a layered embodied large model architecture (integration of planning, decision-making, and execution), eliminating reliance on OpenAI and independently developing end-to-end AI.
Main Achievements
Factory Testing: Successfully completed assembly and handling tasks through 24/7 testing at BMW factories.
Commercial Progress: Supply agreements have been reached with large enterprises, with an expected output of 2,000 units by April 2025 and mass production of 20,000 units in Q3.
Financing Progress: Series B financing of $675 million in 2024, with a valuation of $2.6 billion; seeking Series C financing in 2025, with a valuation of approximately $40 billion.
Future Plans: Deploy 100,000 robots within four years, catering to both industrial and home scenarios.
(3) Unitree
Unitree was founded in 2016 in Hangzhou, China, as a high-tech enterprise focused on the research and development of high-performance quadrupedal robots and general humanoid robots. It is known for its self-developed core components and motion control algorithms, covering consumer, industrial, and performance application scenarios, promoting robot commercialization with a "low-cost, high-performance" strategy. Its quadrupedal robots hold over 60% of the global market share and have achieved mass production of humanoid robots ahead of others.
Development History
2016: The company was established in Hangzhou, with a founding team from Zhejiang University and Huawei, initially focusing on the development of motion control algorithms for quadrupedal robots.
2017: Released the first consumer-grade quadrupedal robot Laikago (weighing 22 kg, with a load capacity of 5 kg), attracting industry attention with a price lower than similar products from Boston Dynamics.
2019-2021: Released industry-grade Aliengo, educational-grade A1, and the bionic robot Go1.
2022: 109 Go1 robotic dogs appeared at the opening ceremony of the Beijing Winter Olympics; launched the industrial-grade B1 and fitness pump PUMP.
2023: Released the upgraded Go2, equipped with 4D LiDAR; introduced the general humanoid robot H1 (the first full-size humanoid robot capable of running in China) and the industrial quadrupedal robot B2.
2024: Completed nearly 1 billion yuan in Series B financing; released the humanoid robot UnitreeG1; selected as one of Forbes China's Top 50 AI Companies.
2025: The humanoid robot H1 "Fu Xi" performed in the CCTV Spring Festival Gala with the act "Yang BOT"; G1 and H1 began limited sales, priced at 99,000 yuan and 650,000 yuan, respectively.
Technical Features
Fully self-developed supply chain: Covers core components such as motors, reducers, controllers, and LiDAR, as well as high-performance motion control algorithms.
High-performance motion capabilities: Go1 achieves natural following interaction, while Go2 is equipped with 4D ultra-wide-angle LiDAR; the B2 industrial robot can run at speeds of up to 6 m/s, adapting to complex terrains; the humanoid robot H1 possesses dynamic balance and rapid running capabilities.
AI Empowerment: Achieves intelligent interaction functions such as dance movement learning and music response through large model training.
Low-cost hardware: Reduces the cost of quadrupedal robots to one-tenth of the industry standard through optimized motor drives, reducer designs, and supply chains.
Main Achievements
Product Matrix: Consumer-grade (Go series), industrial-grade (B series), humanoid robots (H1/G1).
Market Performance: Quadrupedal robots hold over 60% of the global market share, with the Go series becoming a popular product in STEM education. The humanoid robots G1 and H1 quickly sold out during pre-sales on JD.com, with overseas orders accounting for 50%.
International Influence: Featured on international stages such as CES, Super Bowl pre-show performances, and the opening ceremony of the Winter Olympics; products sold to multiple countries worldwide.
Technical Breakthroughs: BeamDojo reinforcement learning allows G1 to complete tasks like walking on balance beams and resisting external disturbances; the HOMIE cockpit system enables precise remote control of the entire body, supporting complex tasks such as dancing and handling.
Commercialization Progress: In 2025, humanoid robots H1/G1 will begin pre-sales, promoting the large-scale application of general robots.
(4) Apptronik
Apptronik is a robotics company headquartered in Austin, Texas, USA, founded in 2016, focusing on the development of general humanoid robots Apollo, aimed at enhancing work efficiency and safety in industries such as manufacturing, logistics, and healthcare through collaboration with humans.
Development History
2016: Apptronik spun off from the Human Centered Robotics Lab at the University of Texas, focusing on the research and development of robotic technology.
2016–2019: Received funding for multiple government and private sector projects (including collaborations with NASA), developing technologies for exoskeletons, humanoid upper bodies, bipedal mobility platforms, and logistics arms.
2020: Developed a humanoid upper body robot.
2022: Developed a fully electric humanoid robot prototype in less than 12 weeks.
August 2023: Released the Apollo Alpha version, marking the first appearance of the Apollo series.
Technical Features
Size and Load Capacity: Apollo stands approximately 173 cm tall, weighs about 73 kg, and can carry approximately 25 kg of items.
Power System: Equipped with a 7-degree-of-freedom robotic arm, with joints in the shoulder, elbow, and wrist for precise object manipulation.
Sensors and Perception Capabilities: Integrated depth cameras and ultra-high-definition long-range LiDAR enhance perception in complex environments.
Safety Design: Utilizes a unique force control architecture to ensure safety when collaborating with humans, similar to collaborative robots rather than traditional industrial robots.
Main Achievements
Industry Applications: Apollo has been tested at Mercedes-Benz's Marienfeld and Kecskemét factories, performing tasks such as parts transportation and quality inspection.
AI Empowerment: Apptronik collaborates with NVIDIA, Google DeepMind, and others to enhance its work capabilities through AI.
Partnerships: Collaborates with supply chain giant Jabil for the production and testing of Apollo robots in its factories, even planning for robots to assist in their own production.
Financing and Support: In February 2025, Apptronik completed a Series A financing of $403 million, with a valuation of approximately $1.5 billion, with investors including Google DeepMind, B Capital, and Capital Factory.
Awards and Honors: Apollo was recognized by Fast Company as a finalist for the 2024 "World Changing Ideas Awards" in the experimental category, acknowledging its potential in addressing labor shortages and industrial automation.
Future Plans: Plans to mass-produce hundreds of units in 2025, with full commercialization in 2026, targeting a price below $50,000.
(5) Boston Dynamics Atlas
Boston Dynamics, a robotics company based in Massachusetts, USA, is known for developing highly mobile and flexible humanoid robots, widely used in industrial automation, logistics, and security.
Development History
1992: The company was founded, initially providing robotic research for the U.S. Army.
2005: Launched the military quadrupedal robot BigDog (load capacity of 154 kg).
2013: Acquired by Google, sold to SoftBank in 2017, and acquired 80% by Hyundai Motor in 2021.
2015: Launched the Spot series, commercialized in 2019, with a global market share of 12.67% in 2023.
2017: Launched Spot Mini, combining wheeled and legged mobility.
2013-2024: Hydraulic-driven version of Atlas (1.5 meters tall, 28 degrees of freedom).
2024: Launched the fully electric version of Atlas, adding neck and waist rotation joints, achieving autonomous handling in factory tests.
Plans for 2025: Pilot production at Hyundai Motor's factory in South Korea.
Technical Advantages
Drive System: Transitioned from hydraulic to fully electric drive, resulting in a more compact structure and more stable control.
Motion Control: Capable of high-speed movement and precise actions in complex environments, with a wide range of waist and neck rotation.
Hand Capabilities: Equipped with a three-finger dexterous hand.
Environmental Perception: 360° perception system, adaptable to complex industrial environments.
AI Collaboration: Jointly developed a large behavior model with Toyota Research Institute for complex decision-making tasks.
Main Achievements
Industrial Testing: In 2024, handling automotive parts at Hyundai Motor's factory.
Leading Global Motion Control Technology: Atlas is regarded as a benchmark for humanoid robot motion control.
Clear Market Positioning: Focused on high-value areas such as high-end industrial inspection and hazardous operations.
Pilot Plans: Initiating mass production pilot in 2025, gradually expanding commercial applications.
4. Current Challenges and Opportunities for Humanoid Robots
Although the humanoid robotics sector has broad prospects, it still faces dual challenges in technology and market. Current technical bottlenecks include high hardware costs, insufficient understanding of the real world by AI models, and precision issues in motion control. Nevertheless, with continuous technological breakthroughs and accelerated capital deployment, this industry still holds significant development potential.
Hardware and Cost Bottlenecks: Tesla's Optimus robot aims to reduce the price of a single unit to $20,000, but current laboratory-level robots still exceed $100,000, leading to slow commercialization progress.
AI and Sensor Technology: Despite rapid advancements in AI, current robots still lack generalization capabilities in complex scenarios, particularly facing significant challenges in real-time perception and decision-making accuracy.
Decentralized Networks: To enable robots to have broader adaptability, decentralized data networks need to possess stronger real-time data processing capabilities.
Innovative solutions based on Web3 are rapidly emerging. New models such as decentralized AI training and robotic economies will significantly reduce development costs while promoting further dissemination of hardware and technology. Just as Web3 has shown potential in the metaverse and AI fields, the humanoid robot sector is expected to become the next core narrative of Web3.
II. Overview of Web3 Robots and Embodied Intelligence Projects
1. Overview
With the rapid development of AI, robotics technology, and blockchain, Web3 robot projects are becoming an important intersection of emerging technologies. Just like every significant historical transformation, which patiently awaited the arrival of each puzzle piece, the skill tree of humanoid robotics technology is now ready to be illuminated. By integrating embodied intelligence, decentralized physical infrastructure networks, token incentive mechanisms, and AI agents, the goal is to create an open, collaborative, self-driven "decentralized machine economy."
General characteristics of Web3 robots and embodied intelligence projects include:
Real-world task execution: Robots collect environmental information through perception modules (such as cameras and LiDAR) and are guided by AI decision-making modules to perform tasks such as navigation, monitoring, and data collection;
Decentralized collaboration mechanisms: Incentivizing robot hardware providers, data contributors, and computing nodes through a token system to achieve resource sharing across regions and entities;
AI Agent and Robot Integration: Using AI agents as the "decision center" (brain) and robots as "embodied executors" (body) to achieve the integration of cognition and action;
Ecological Subfields: Current Web3 robot competitions can be divided into the following eight directions:
Robot hardware platforms
Robot software and protocol layers
Decentralized positioning and data networks
Spatial intelligence and environmental modeling
Infrastructure and underlying networks (Layer-1)
Machine economy systems
DePAI DAO organizational forms
Gamified/community-based robot projects
2. Current Web3 Robots and Embodied Intelligence Project Introductions
(1) BitRobot
Project Overview
BitRobot Network is a decentralized network based on a subnet architecture, aimed at accelerating the development of embodied artificial intelligence (Embodied AI) through cryptoeconomic incentives. Its core goal is to solve the current data and resource bottlenecks faced in the Embodied AI field by aggregating computing resources, robot clusters, datasets (real or synthetic), and AI models, thereby promoting large-scale innovation in robotics technology.
Financing Situation
At the beginning of 2025, FrodoBots Lab completed a $6 million seed round of financing, collaborating with Protocol Labs to develop BitRobot. The round was led by Protocol VC, with participation from Big Brain Holdings, Fabric Ventures, Solana Ventures, and co-founders of Solana and several DePIN projects.
Product Logic
BitRobot can be simply understood as a "robot collaboration network" incentivized by blockchain and tokens, allowing everyone to share robots, computing power, and data to accelerate the development of AI robots together.
Subnet Identities
Subnet Owners: Define subnet tasks (VRW) and allocate reward ratios.
Subnet Validators: Validate the effectiveness of contributor outputs.
Subnet Contributors: Provide resources such as robot hardware, computing power, and manpower.
Economic Model and Incentive Mechanism
Network Rewards: Incentivize resource providers, such as robot owners and computing power suppliers, through tokens.
Resource Utilization: Third parties, such as AI laboratories, pay fees to use subnet resources, such as testing models with robot clusters.
Commercialization: Open datasets and models for non-commercial use, with commercial licensing revenue reinvesting into the ecosystem.
(2) Reborn
Project Overview
Reborn transforms human motion data into token assets to build a Robot Foundation Model (RFM), thereby training humanoid robots with general capabilities.
Ecological Product Matrix
Unified Data Platform: A distributed infrastructure that collects high-quality real-world and synthetic data through motion capture wearables, VR interactions, mobile video, and the Roboverse simulation engine, laying the foundation for scalable training.
Open Model Ecosystem: A reusable library of embodied AI models, including OpenVLA (visual-language-action model), full-body controllers, and dexterous manipulation models, allowing developers to deploy or extend directly through the Reborn physical AI application store.
Physical Agents: Robotic agents in the real world, planned to evolve through three stages: collaborative teleoperation, dedicated model deployment, and general autonomy, forming a data-driven practical path.
Network Roles
Data Contributors: Community users contribute VR/AR game data and real body motion data to train the RFM model while earning rewards.
Network Validators: Validators ensure the authenticity and usability of the data and receive rewards.
Data Demanders: Robot developers purchase model usage rights or data.
Hardware Devices
According to the project’s official website, the project has sold over 8,000 Rebocap™ devices, which are used for posture data capture and data collection.
Data Categories
Embodied Vlog (real-world task videos): Videos captured from a first-person perspective using common cameras, such as smartphones and GoPros, showcasing fine manipulation tasks like making sandwiches or washing dishes.
Mocap Life (precise motion capture using Rebocap™ devices): Collecting precise human joint data.
VR Games: Capturing hand markers and interaction operation data through partnerships with VR platforms.
Financing Situation
According to the official website, the project has received support from GGV Capital, but specific financing details have not been disclosed.
(3) Vana
Project Overview
Vana is a decentralized protocol aimed at achieving data sovereignty and a user-owned AI data layer. It allows users to own, control, and monetize their personal data while building a DataDAO ecosystem. Vana views data as a new type of digital asset, supporting users in aggregating data into the DataDAO for training AI models, thereby breaking the data monopoly of large tech companies and promoting the development of an open AI economy. Vana has attracted over a million users to form the DataDAO, covering fields such as health, robotics, science, and prediction markets.
Financing Situation
Vana has completed a total of $25 million in financing, including a seed round led by Polychain Capital and an A round led by Paradigm, with participation from Coinbase Ventures. This financing is used to accelerate the development of user-owned AI and ecosystem expansion, supporting incubated projects within the DataDAO.
Business Logic
Vana's product logic revolves around shifting data from "extracted resources" to "user-controlled assets." Users upload encrypted data to Vana's digital wallet, aggregating data through the DataDAO to form a Data Liquidity Pool (DLP). AI developers access these pooled data through smart contracts for model training, while data contributors receive governance tokens and rewards. This logic emphasizes privacy protection, collective negotiation, and market-driven value discovery, avoiding the public verification issues of traditional blockchains.
Technical Advantages
Privacy Protection Technology: Encrypted data ownership records, access control, and verification proofs, supporting ZK technology to ensure privacy.
Strong Network Structure Scalability: Global state maintenance of data ownership, access rights, and token balances; natively supports DataDAO pooled data; VRC-20 token standard used for governance.
Interoperability: ERC20 wrapped versions support the Ethereum ecosystem; integration of infrastructure services like Ser optimizes data access pipelines.
Robot Narrative
Vana positions itself as a key data infrastructure within the robot network, providing user-owned real-world video and sensor data for training physical AI and autonomous systems. The project emphasizes that permissioned user data can scale to support the development of robotic technology, avoiding reliance on platform monopolies. Through the DataDAO, Vana aggregates diverse human data to facilitate better interactions between robots and the world. Currently, Vana supports the incubation of projects within the robotic ecosystem, such as leading projects like PrismaX.
Token Situation
The project is set for TGE in December 2024, with a current circulating market cap of $134M and a total market cap of $525M, and it is listed on major exchanges like Binance and Upbit.
(4) PrismaX
Project Overview
PrismaX is a decentralized data marketplace platform focused on providing fuel for real-world robot development. Through community-driven data collection and incentive mechanisms, it bridges the gap between robots and mainstream adoption. The project develops multimodal generative AI to help robots "see, understand, and interact" with the physical world through visual, video, and sensor data, rather than being limited to text data, aiming to solve the data bottleneck for robots and achieve autonomous and efficient coordination.
Business Model: Data Collection - Remote Operation - Model Training
B2B Data Services: Providing high-quality, diverse visual datasets to robot/AI companies.
Remote Operation (Teleop) Platform: Offering standardized remote control solutions, including operator management, payment, and software interfaces, charging based on usage or subscription.
Model Training and API Services: Collaborating with AI companies to provide pre-trained models or data augmentation services, charging licensing fees or revenue sharing.
Decentralized Data Economy: Incentivizing data contributors through crypto and taking a cut from data transactions.
Roadmap
Phase I: Remote Operation + Data Collection (2025-2026): Core goal: Establish a global distributed network of remote operators, with robots primarily used for AI training data collection.
Phase II: Real Tasks + Edge Models (2026-2027): Core goal: Robots perform real commercial tasks such as logistics and manufacturing, with operators managing multiple robots and AI assistance to reduce latency.
Phase III: Fully Autonomous + Machine Economy (2027+): Core goal: Achieve a highly automated robot service network, with foundational models driving autonomous decision-making.
Financing Situation
In June 2025, completed $11 million in financing, led by a16z CSX, with participation from Virtuals Protocol, Volt Capital, Symbolic Capital, Stanford Blockchain Accelerator, and others.
(5) OpenMind
Project Overview
OpenMind's OM1 is a modular AI runtime environment designed for agents and robots, featuring multiple functionalities such as movement and voice.
Key Features
OM1 supports the deployment of AI agents in both digital and physical worlds after configuration. Simply create an AI agent to run in the cloud or execute on physical robot hardware—including quadrupedal robots, with upcoming support for TurtleBot 3 and humanoid robots.
Application Scenarios
AI agents built on OM1 can integrate multi-source data (internet, X/Twitter, cameras, LiDAR), enabling actions such as tweeting, exploring your room, or shaking hands and conversing with you. Through OM1, you can talk to OpenAI's gpt-4o and even "shake hands" with it.
Technical Characteristics
Multi-Agent Endpoints: Allowing different types of AI agents to collaborate through a single API.
Planner: Responsible for decision-making and task decomposition, determining goals and priorities.
Navigator: Responsible for formulating execution paths and optimizing task flows.
Perception/Control: Handling environmental perception and specific operational execution.
Modularity and Scalability: The system employs a modular design, allowing developers to add or replace agents as needed. This flexibility makes it suitable for various application scenarios, from robot control to complex decision-making systems.
Open Source Support
OpenMind provides some open-source code on GitHub to demonstrate how to generate solutions through recursive prompts and integrate with OpenAI's GPT-4o API.
(6) FrodoBots
Project Overview
An innovative project focused on artificial intelligence and robotics, dedicated to building an AI-driven humanoid robot fighting club and a global robot network. Through DePIN technology, it creates an ecosystem that combines robot competitions, gaming, and research.
Financing Situation
It is reported that FrodoBots secured $2 million in pre-seed financing at the project's inception; in early 2025, FrodoBots Lab completed a $6 million seed round of financing in collaboration with Protocol Labs to develop BitRobot.
Core Features
Robot Fighting Club: Developing humanoid robots for real-time competitions, merging AI with entertainment.
Global Robot Network: Establishing a distributed fleet of robots for data collection, urban navigation research, and other applications.
AI and Dataset Innovation: Generating unique datasets through robot competitions to advance AI algorithms and robotics technology.
Community and Collaboration: Co-hosting AI workshops and demonstration events with global partners to promote technology exchange and application implementation.
Economic Model and Token
The project recently launched the robot meme token $SAM @SamIsMoving, a robot that collects real-time data and periodically posts street scene information on its official X account.
(7) XMAQUINA
Project Overview
XMAQUINA is a DePAI DAO that provides decentralized investment opportunities in robotics and machine finance. Holding the platform token $DEUS grants access to investment opportunities and robot income.
Main Vision
Building a Decentralized Machine Economy: Creating a community-driven ecosystem that achieves decentralized management and collaboration of machines and robotic devices through cooperation with the peaq network.
On-Chain Robot Management Platform: Providing registration, certification, and data management for robotic devices, ensuring trustworthy operations within the decentralized network.
Community-Driven Governance and Distribution: Empowering the community to collectively decide the rules and resource allocation of the machine economy through the $DEUS token and DAO mechanism.
Technology and Applications
Marketplace: Offering a decentralized marketplace for trading robot components, AI models, and services, promoting innovation and technology sharing.
Deus Labs: The R&D department within the DAO, focusing on the development of open-source robotics technology, ethical AI, and decentralized technologies.
AI Agents: Developing AI agents to automate the operations and governance processes of the DAO.
Financing Situation
At the end of April 2025, 25 million $DEUS tokens were sold in the IDO, accounting for 2.5% of the total supply (1 billion $DEUS), with each $DEUS token priced at $0.04.
In February 2025, community fundraising was conducted through the DEUS Genesis Auction, successfully raising 3,703,703 $PEAQ tokens in the first Genesis Auction on the peaq network.
It is reported that the project completed a seed round of financing early on, with participation from institutions such as EoT Ventures and Moonrock Capital, though the specific amount has not been disclosed.
(8) GEODNET
Project Overview
A decentralized real-time dynamic positioning network providing centimeter-level GPS accuracy for robots, drones, and automated machines. It builds an AI-driven navigation system and serves as a geographic information center for the real world.
How It Works
Base Stations: Individuals or businesses purchase and install GNSS hardware devices (satellite miners) to collect satellite positioning signals in locations with a clear view of the sky and upload data via the internet. These base stations calculate signal transmission time differences through models to generate high-precision RTK correction data.
Blockchain and Token Incentives: Using the $GEOD token to incentivize participants. Base station operators earn GEOD token rewards by providing high-quality GNSS data, while data users access the network by purchasing RTK correction data, which reduces token circulation to support token value.
Decentralized Verification: Employing Proof of Accuracy and Proof of Stake protocols to ensure data accuracy and network security, preventing data tampering.
Financing Situation
In February 2025, completed $8 million in financing (OTC), led by Multicoin Capital, with participation from ParaFi Capital, Digital Asset Capital Management, and others.
In April 2024, completed $2 million in financing, with participation from Pantera Capital, CoinFund, VanEck, Santiago Roel Santos, and others.
Token Situation
The project is set for TGE in Q3 2023, with a current circulating market cap of $57M and a total market cap of $179M, and it is listed on exchanges such as Gate and MEXC.
(9) Auki/Posemesh
Project Overview
A DePIN-based spatial computing project aimed at building a decentralized machine perception network. The project enables devices to securely and privately share spatial data and computing power through a collaborative spatial computing protocol, forming a shared understanding of the physical world.
Business Logic
Data Input: Any device equipped with sensors can participate in the Posemesh network to provide spatial data, with nodes offering computing power and storage.
Data Processing: The Posemesh network integrates data to generate high-precision spatial maps and services.
Data Output: Businesses/developers use the data to optimize operations (such as robot navigation, AR experiences) and pay with tokens.
Ecological Incentives: Data providers and node operators receive token rewards, while Auki Labs profits from transaction fees and customized services.
Positive Cycle: The token economy incentivizes more devices and nodes to join, expanding the network scale and further enhancing data accuracy and application scenarios.
Financing Situation
The project conducted a community whitelist presale around August 2024, with 390 qualified addresses participating according to the official blog, raising approximately 204,700 USDT.
Token Situation
The project had its TGE at the end of August 2024, with a current circulating market cap of 69M and a total market cap of 284M, and it is listed on the MEXC exchange.
(10) peaq
Project Overview
peaq is a Layer 1 focused on DePIN and machine economy, established in 2017.
Project Features
DePIN Infrastructure: peaq provides modular DePIN functionalities, supporting developers in building decentralized physical infrastructure projects across various fields, including IoT, mobility, energy, agriculture, and gaming.
Machine Economy: By assigning autonomous identities, payment capabilities, and access control to machines, devices can independently conduct transactions, data exchanges, and collaborations.
Cross-Chain Interaction: Interconnected with over 90 blockchains, supporting cross-chain data and liquidity flow.
$PEAQ Token: As the native token of the network, $PEAQ is used for transaction fees, network governance, machine rewards, NFT ownership, and DeFi liquidity provision.
Key Functions
Decentralized Identity and Control (peaq ID and peaq Control): Assigning a self-managed digital identity (peaq ID) to each device for identification and authentication. peaq Control is a Web3 machine control center that connects devices, sensors, vehicles, and robots, ensuring secure interaction with the network.
Access Control and Permission Management (peaq Access): A role-based access control mechanism that allows developers to define user groups, roles, and permissions to control access to devices or services. For example, in IoT applications, only authorized users can access shared vehicles.
Decentralized Payment and Economic Model (peaq pay): Supporting peer-to-peer payments between machines and between machines and users, ensuring fund security and transaction transparency. peaq pay uses multi-signature wallets to verify fund sufficiency and processes refunds or fee settlements after service completion.
Data Verification and Trustworthiness (peaq verify): Ensuring that data in the DePIN network is authentic and sourced from actual physical devices, preventing fraud and data tampering. Verification is divided into three layers: machine origin authentication, data consistency checks, and community verification.
Modular DePIN Functions and Developer Support: Providing a one-stop technology stack, including smart contract support, cross-chain bridging, data storage, and AI agents.
Ecosystem Incentives (DePIN Flywheel): Designing incentive mechanisms through the "DePIN Flywheel" model, where participants (device owners, developers, users) earn token rewards for contributing resources or data, attracting more devices to join the network and creating a positive cycle.
Financing Situation
In May 2024, completed a $20 million fundraising activity through CoinList, with participation from over 14,500 community members.
In March 2024, completed $15 million in financing, led by Generative Ventures and Borderless Capital, with participation from several institutions including Spartan Group, HV Capital, CMCC Global, and Animoca Brands.
Token Situation
The project will have its TGE in November 2024, with a current circulating market cap of 135M and a total market cap of 483M, and it is listed on exchanges such as Gate, Bitget, and MEXC.
(11) CodecFlow
Project Overview
CodecFlow is an AI operator and robot execution layer based on the Solana blockchain, focused on addressing the limitations of traditional automation in software and robotics. It enables AI to perceive, reason, and execute actions in screen or robotic systems through the Vision-Language Actions (VLA) model. The project includes Fabric (a multi-cloud execution layer for GPU-intensive workloads) and Operator Kit (optr) (a lightweight toolkit for building desktop, browser, simulation, and robotic agents).
Product Logic
The product logic of CodecFlow revolves around the end-to-end execution of AI operators: from perception (visual-language input) to reasoning (VLA model processing) to action (cross-screen/robot execution). Traditional automation relies on fragile UI scraping or APIs, while CodecFlow dynamically optimizes resource allocation using Fabric, selecting the lowest-cost nodes to execute GPU tasks; it modularly manages agent states, actions, and security, ensuring composability and safety. This logic emphasizes the robustness of distributed computing, supporting developers in seamless transitions from local to remote, suitable for software automation and physical robot interaction.
Core Features
VLA Model Integration: AI agents perceive the environment, reason decisions, and execute tasks through the visual-language action framework, supporting screen interaction and robot control.
Fabric Execution Layer: Multi-cloud GPU scheduling, real-time sampling capacity and pricing, selecting optimal nodes to run compute-intensive workloads.
Optr Toolkit: A lightweight SDK for building agents, providing action management, state tracking, simulation running, and security modules, supporting desktop/browser/robot environments.
Remote Simulation Support: Developers can run simulations via servers without needing high-end local hardware, enhancing accessibility.
Token Situation
The project will have its TGE in May 2025, with a current circulating market cap of 30M and a total market cap of 40M, not yet listed on exchanges, primarily traded on DEX.
CA:69LjZUUzxj3Cb3Fxeo1X4QpYEQTboApkhXTysPpbpump
(12) RICE AI
Project Overview
RICE AI focuses on AGI robot development. It builds a DePIN network based on the BNB Chain, connecting global robots for data sharing and collaboration, addressing the data scarcity issue in robot AI development. The project is supported by the Hong Kong robotics company Rice Robotics, with the deployed Minibot M1 robot as its core hardware foundation. The platform aims to collect high-quality real-world robot training data through crowdsourcing, promoting the democratization of the robotics industry and accelerating AGI development.
Financing Situation
The project will conduct a presale of the $RICE token through the TokenFi Launchpad in August 2025, raising $750,000, accounting for 10% of the total supply, with a project FDV of $7.5 million.
Product Logic
The product logic of RICE AI is built around the DePIN model: global robots (including Minibot M1) upload sensor and interaction data through the network in exchange for $RICE token rewards; this data enters a decentralized AI minting factory for training foundational robot models. Developers can subscribe to models for AI application development, receiving discounts for using $RICE. The platform monetizes data through tokenization, addressing the high costs, delays, and silos of robot data collection, promoting the transition from simulated data to real-world data, ultimately achieving robot autonomous learning and collaboration.
Core Features
Data Contribution and Rewards: Robot users upload interaction data (such as object grasping, movement trajectories) through VR remote control or automatic collection, earning $RICE.
Model Training and Access: The decentralized minting factory uses data to train physical AI foundational models, supporting subscription access (with discounts for $RICE payments).
Governance and Data Market: $RICE is used for platform governance and data trading, with a portion of platform fees allocated for token buyback and burn.
Robot Integration: Supporting hardware like Minibot M1, integrating GPU processing for real-time learning and collaboration.
Multi-Source Data Support: Including third-party robots, remote control operations, self-centered videos, etc.
Token Situation
The project will have its TGE in August 2025, with a current circulating market cap of 22M and a total market cap of 120M, listed on Binance Alpha and set to be listed on MEXC.
### Summary and Outlook
Currently, the Web3 humanoid robot and embodied intelligence field is still in its early stages, with the overall ecosystem yet to take shape, and industry standards and application paradigms continuously evolving. From the observed forms of Web3 humanoid robot projects, they can mainly be divided into three categories:
First, new projects that focus on Web3 humanoid robots from the outset, which have no historical baggage and can integrate on-chain incentive mechanisms with embodied intelligence in a more native logic, offering significant imaginative potential.
Second, projects originally focused on DePIN, AI, or data that have recently expanded their narrative into the robotics track. These projects often have more mature teams and some community accumulation, providing a foundation for rapid validation and implementation.
Third, lightweight projects primarily centered around memes, gaming, and social interaction, which, while having lower technical barriers, possess strong community operation capabilities and narrative flexibility, potentially leading to the early emergence of blockbuster products.
In summary, the current challenges include technology integration, interaction experience, and user education, but the paradigm shift brought about by the combination of embodied intelligence and Web3 still holds immense potential. Humanoid robots, as the most tangible carriers of embodied intelligence, once deeply integrated with on-chain identities, incentive mechanisms, and decentralized autonomous systems, are expected to create truly on-chain intelligent agents that can not only participate in the digital economy but also influence the production and collaboration methods in the real world. We firmly believe that Web3 humanoid robots will become an important narrative direction worthy of long-term attention and may serve as a key entry point connecting the physical world and virtual space.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
