Web3 Robot Research Report: The Current Status and Outlook of Decentralized Machine Economy and Embodied Intelligence

Original author: https://x.com/snow949494

Abstract

This report explores how the technology trends dominated by traditional tech giants are migrating to the cryptocurrency market and predicts that this cyclical pattern will reappear in the humanoid robot 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 faced by top humanoid robot 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 cutting-edge insights for industry participants, revealing the strategic significance of a decentralized machine economy in connecting the physical and digital worlds.

I. 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 robot sector

1. Review of the migration from tech giants to the mainstream narrative of Crypto

The narrative evolution of Web3 is heavily influenced by the movements of traditional tech giants and financial capital. Looking back at the past few rounds of 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 towards the metaverse, investing tens of billions of dollars in developing 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) rising 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, its astonishing gains remain a hallmark of crypto history.

AI wave (2023--2024):

• At the end of 2022, OpenAI launched ChatGPT, igniting a global AI craze, marking the arrival of the generative artificial intelligence era. Microsoft quickly responded, 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, reaching over 10 times their value by early 2024, setting historical highs. Decentralized AI projects like Bittensor (TAO) attempted to build decentralized neural network model training and incentive mechanisms, with their token prices leading mainstream assets to yield over 10 times excess returns in the following period. By the end of 2024, the market was driven by the AI Agent and related infrastructure, giving rise to several hundred-fold opportunities in a short time, 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 capital firms towards humanoid robots is rapidly increasing. This "tech trend --- capital influx --- narrative migration" chain reaction model is being replayed, with the focus shifting 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 about 75% of jobs in the U.S. (63 million positions) are "humanoid robot adaptable," potentially affecting wage expenditures 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 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

Goldman Sachs pointed out in its February 2025 report "Humanoid Robots III" 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), Unitree Technology (China), AGIBOT (China), and Leju Robotics (China) have 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, CEO Elon Musk stated that humanoid robots will become the industrial main force, with numbers expected to exceed 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-purpose 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); around 10 years, general-purpose 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 at the 2025 GTC conference the launch of Isaac GR00T N1, the first open-source robotic foundational model platform, using 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 Blue entertainment robot indicates that robots are no longer limited to industrial scenarios but are entering the consumer and interactive experience fields.

3. Introduction to global top humanoid robot companies

(1) Tesla Optimus

Tesla's general-purpose 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, targeting a price below $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-metal version was showcased at AI Day.

• March-December 2023: The robot achieved walking, picking up objects, visual recognition, simple actions (like 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 automatic 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 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 22 degrees of freedom dexterous hands.

• 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: Uses the FSD computing platform, supporting autonomous navigation and environmental perception.

• Execution structure: 14 rotary joints (frameless motors + harmonic reducers), 14 linear joints (planetary roller screws), 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 deployed in Tesla factories starting in 2024.

• Market planning: Thousands of units expected to be produced in 2025, with an estimated 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, Figure AI is a U.S. robotics company focused on developing AI-driven humanoid robots aimed at addressing labor shortages, particularly in manufacturing and retail.

Development history

• May 2022: The company was established, focusing on humanoid robot research and development.

• October 2023: Released Figure 01, integrating GPT-4 for natural language interaction and object classification.

• February 2024: Completed a Series B funding round of $675 million (with investments from Microsoft, Nvidia, OpenAI, etc.), valuing the company at $2.6 billion.

• August 2024: Released Figure 02, with comprehensive hardware upgrades, featuring 16 degrees of freedom dexterous hands and a 20-hour battery life.

• November 2024: 40 units of production demand, expected to reach 2,000 units by April 2025, with a target of 20,000 units in Q3.

• February 2025: Figure AI is seeking $1.5 billion in Series C funding, with a valuation of approximately $40 billion.

• Future plans: Aim to ship 100,000 units within 4 years, with a new production base set to open in 2025.

Technical features

• Vision and perception: Equipped with 6 RGB cameras and LiDAR, capable of comprehensive environmental perception.

• Hand capabilities: Single hand with 16 degrees of freedom, capable of carrying 25 kg, with high precision.

• Computing platform: Equipped with NVIDIA RTX GPU, with inference capabilities three times that of the previous generation.

• AI system: Adopts a layered embodied large model architecture (planning, decision-making, execution integration), eliminating reliance on OpenAI and independently developing end-to-end AI.

Main achievements

• Factory testing: Successfully completed assembly and transportation tasks through 24/7 testing at BMW factories.

• Commercial progress: Agreements reached with large enterprises for supply, with 2,000 units expected to be produced by April 2025 and 20,000 units in mass production in Q3.

• Funding progress: $675 million raised in Series B funding in 2024, with a valuation of $2.6 billion; seeking Series C funding in 2025, with an estimated valuation of approximately $40 billion.

• Future plans: Deploy 100,000 robots within 4 years, covering both industrial and home scenarios.

(3) Unitree

Founded in 2016 in Hangzhou, China, Unitree is a high-tech enterprise focused on the research and development of high-performance quadrupedal robots and general-purpose humanoid robots, known for its fully self-developed core components and motion control algorithms. Its products cover consumer, industrial, and performance application scenarios, promoting robot commercialization with a "low-cost, high-performance" strategy, with over 60% global market share in quadrupedal robots and being the first to achieve mass production of humanoid robots.

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, 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 robot dogs appeared at the opening ceremony of the Beijing Winter Olympics; launched industrial-grade B1 and fitness pump PUMP.

• 2023: Released the upgraded Go2, equipped with 4D LiDAR; general-purpose humanoid robot H1 (the first full-size humanoid robot capable of running in China) and industrial quadrupedal robot B2.

• 2024: Completed nearly 1 billion yuan in Series B financing; released humanoid robot UnitreeG1; selected as one of Forbes China's Top 50 AI companies.

• 2025: Humanoid robot H1 "Fu Xi" appeared at the CCTV Spring Festival Gala performance "Yang BOT"; G1 and H1 began limited sales, priced at 99,000 yuan and 650,000 yuan, respectively.

Technical features

• Fully self-developed across the entire industry 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, Go2 is equipped with 4D ultra-wide-angle LiDAR; B2 industrial robot runs at a speed of 6 m/s, adapting to complex terrains; humanoid robot H1 possesses dynamic balance and rapid running capabilities.

• AI empowerment: Achieves intelligent interactive functions such as dance action learning and music response through large model training.

• Low-cost hardware: By optimizing motor drives, reducer designs, and supply chains, the cost of quadrupedal robots has been reduced to one-tenth of the industry average.

Main achievements

• Product matrix: Consumer-grade (Go series), industrial-grade (B series), humanoid robots (H1/G1).

• Market performance: Over 60% global market share for quadrupedal robots, with the Go series becoming a popular product in STEM education. 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, and the opening ceremony of the Winter Olympics; products sold to multiple countries worldwide.

• Technical breakthroughs: BeamDojo reinforcement learning enables G1 to complete tasks like walking on balance beams and resisting external disturbances; HOMIE cockpit system allows for precise remote control of the entire body, supporting complex tasks such as dancing and transportation.

• Commercial progress: Humanoid robots H1/G1 began pre-sales in 2025, promoting large-scale applications of general-purpose robots.

(4) Apptronik

Apptronik is a robotics company headquartered in Austin, Texas, USA, founded in 2016, focusing on developing general-purpose humanoid robots, Apollo, aimed at improving work efficiency and safety in industrial, logistics, and medical fields 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 exoskeletons, humanoid upper bodies, bipedal mobile platforms, and logistics arms.

• 2020: Developed a humanoid upper body robot.

• 2022: Developed a fully electric humanoid robot prototype, with a development cycle of 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 is approximately 173 cm tall, weighs about 73 kg, and can carry about 25 kg of items.

• Power system: Equipped with a 7-degree-of-freedom robotic arm, with joints at the shoulder, elbow, and wrist, allowing for precise manipulation of objects.

• Sensors and perception capabilities: Integrated depth cameras and ultra-high-definition long-range LiDAR, enhancing perception capabilities in complex environments.

• Safety design: Adopts a unique force control architecture, ensuring safety during collaboration 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 working capabilities through AI.

• Partnerships: Collaborated with supply chain giant Jabil for the production and testing of Apollo robots in its factories, even planning for robots to assist in its own production.

• Financing and support: In February 2025, Apptronik completed a $403 million Series A funding round, with an estimated 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 a target price below $50,000.

(5) Boston Dynamics Atlas

A robotics company based in Massachusetts, USA, 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% stake 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 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.

Technical advantages

• Drive system: Transitioned from hydraulic to fully electric, with a more compact structure and 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 large behavioral models 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-end industrial inspection and hazardous operations in high-value areas.

• Pilot plans: Initiating mass production trials in 2025, gradually expanding commercial applications.

4. Current challenges and opportunities for humanoid robots

Although the humanoid robot sector has broad prospects, it still faces dual challenges of 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 AI development, current robots still lack generalization capabilities in complex scenarios, particularly in real-time and precise perception and decision-making.

• Decentralized networks: To enable robots to have broader adaptability, decentralized data networks need stronger real-time data processing capabilities.

Innovative solutions based on Web3 are rapidly emerging. New models such as decentralized AI training and the robotic economy will significantly reduce development costs while promoting further hardware and technology adoption. 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. Like every significant historical transformation, it is patiently waiting for each piece of the puzzle to arrive, and now the skill tree of humanoid robots can be illuminated. 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 features of Web3 robots and embodied intelligence projects include:

• Execution of real-world tasks: Robots collect environmental information through perception modules (such as cameras, LiDAR) and are guided by AI decision-making modules to complete tasks such as navigation, monitoring, and data collection;

• Decentralized collaboration mechanisms: Incentivizing robot hardware providers, data contributors, and computing nodes through a token system, achieving resource sharing across regions and entities;

• Integration of AI agents and robots: Using AI agents as the "decision-making center" (brain) and robots as the "embodied executors" (body), achieving integration of cognition and action;

• Ecological segmentation: Current Web3 robot events can be divided into the following 8 directions:

1. Robot hardware platforms

2. Robot software and protocol layers

3. Decentralized positioning and data networks

4. Spatial intelligence and environmental modeling

5. Infrastructure and underlying networks (Layer-1)

6. Machine economy systems

7. DePAI DAO organizational forms

8. Gamified/community-based robot projects

2. Current Web3 robots and embodied intelligence project introductions

(1) BitRobot

https://x.com/BitRobotNetwork

Project Overview

BitRobot Network is a decentralized network based on a subnet architecture, aiming to accelerate the development of embodied AI through cryptoeconomic incentives. Its core goal is to aggregate computing resources, robot clusters, datasets (real or synthetic), and AI models to address the current data and resource bottlenecks faced in the field of embodied AI, promoting large-scale innovation in robotics technology.

Financing Situation

In early 2025, FrodoBots Lab completed a $6 million seed round of financing, collaborating with Protocol Labs to develop BitRobot. Led by Protocol VC, with participation from Big Brain Holdings, Fabric Ventures, Solana Ventures, and others, including co-founders of Solana and several DePIN project founders.

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, accelerating the development of AI robots together.

Subnet Roles

• 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 human resources.

Economic Model and Incentive Mechanism

• Network rewards: Token incentives for resource providers, such as robot owners and computing power suppliers.

• Resource utilization: Third parties such as AI labs pay fees to use subnet resources, such as testing models with robot clusters.

• Commercialization: Non-commercial use of open datasets and models, with commercial licensing revenue feeding back into the ecosystem.

(2) Reborn

https://x.com/reborn_agi

Project Overview

Reborn transforms human movement data into token assets to build robotic foundational models (RFM), thereby training humanoid robots with general capabilities.

Ecological Product Matrix

1. Unified Data Platform: A distributed infrastructure that collects high-quality real-world and synthetic data through motion capture wearables, VR interactions, mobile video, and Roboverse simulation engines, laying the foundation for scalable training.

2. Open Model Ecosystem: A reusable library of embodied AI models, including OpenVLA (visual-language-action models), full-body controllers, and dexterous manipulation models, allowing developers to deploy or extend directly through the Reborn physical AI application store.

3. 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

1. Data Contributors: Community users contribute VR/AR game data and real-body motion data to train RFM models while earning rewards.

2. Network Validators: Validators ensure the authenticity and usability of the data, earning rewards.

3. Data Demanders: Robot developers purchase model usage rights or data.

Hardware Devices

According to the project’s official website, there are over 8,000 Rebocap™ units sold, used for posture data capture and data collection.

Data Categories

1. Embodied Vlog (real-world task videos): Captured using common cameras, such as smartphones and GoPros, showcasing detailed operation tasks like making sandwiches or washing dishes.

2. Mocap Life (precise motion capture using Rebocap™ devices): Collects precise human joint data.

3. VR Games: Captures 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, with specific financing details yet to be disclosed.

(3) Vana

https://x.com/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 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 DataDAO, covering health, robotics, science, prediction markets, and other fields.

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 and others. This financing is used to accelerate the development of user-owned AI and ecosystem expansion, supporting incubated projects within 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 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 scalability of network structure: 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 foundational services like Ser optimizes data access pipelines.

Robot Narrative

Vana positions itself as a key data infrastructure in 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 robotics technology, avoiding reliance on platform monopolies. Through DataDAO, Vana aggregates diverse human data to facilitate better interaction between robots and the world. Currently, Vana supports the incubation of projects within the robotics ecosystem, such as leading projects like PrismaX.

Token Situation

The project had its TGE in December 2024, currently with a circulating market cap of $134M, total market cap of $525M, and listed on major exchanges like Binance and Upbit.

(4) PrismaX

https://x.com/PrismaXai

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

1. 2B data services: Providing high-quality, diverse visual datasets to robot/AI companies.

2. Remote operation (Teleop) platform: Offering standardized remote control solutions, including operator management, payment, and software interfaces, charging based on usage or subscription.

3. Model training and API services: Collaborating with AI companies to provide pre-trained models or data augmentation services, charging licensing fees or revenue sharing.

4. Decentralized data economy: Incentivizing data contributors through crypto and taking a cut from data transactions.

Roadmap

1. Phase I: Remote operation + data collection (2025-2026): Core goal: Establish a globally distributed remote operator network, with robots primarily used for AI training data collection.

2. 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 assisting in reducing latency.

3. 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

https://x.com/openmind_agi

Project Overview

OpenMind's OM1 is a modular AI runtime environment for agents and robots, equipped with multiple functions such as motion and voice.

Main Features

OM1 supports deploying AI agents into 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, and soon supporting TurtleBot 3 and humanoid robots.

Application Scenarios

AI agents built on OM1 can integrate multi-source data (web, X/Twitter, cameras, LiDAR), enabling them to tweet, explore your room, or shake hands or converse with you. Through OM1, you can talk to OpenAI's gpt-4o and even "shake hands" with it.

Technical Features

• 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 operation execution.

• Modularity and scalability: The system employs a modular design, allowing developers to add or replace agents based on needs. This flexibility makes it suitable for various application scenarios, from robot control to complex decision 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

https://x.com/frodobots

Project Overview

FrodoBots is an innovative project focused on artificial intelligence and robotics technology, dedicated to building an AI-driven humanoid robot fighting club and a global robot network. Through DePIN technology, it creates an ecosystem combining robot competitions, gaming, and research.

Financing Situation

FrodoBots reportedly secured $2 million in pre-seed funding at the project's inception; in early 2025, FrodoBots Lab completed a $6 million seed round of financing, collaborating with Protocol Labs to develop BitRobot.

Core Functions

• Robot fighting club: Developing humanoid robots for real-time competitions, integrating AI and 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 promote the development of AI algorithms and robotics technology.

• Community and collaboration: Jointly hosting AI workshops and demonstration events with global partners to promote technical exchange and application landing.

Economic Model and Token

The project recently launched the robot meme token $SAM @SamIsMoving, a robot that collects real-time street data and periodically posts street scene information on its official X account.

(7) XMAQUINA

https://x.com/xmaquinaDAO

Project Overview

XMAQUINA is a DePAI DAO providing 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 partnerships with the peaq network.

• On-chain robot management platform: Providing registration, certification, and data management for robotic devices, ensuring trustworthy operations in a decentralized network.

• Community-driven governance and distribution: Empowering the community to jointly decide the rules and resource allocation of the machine economy through the $DEUS token and DAO mechanism.

Technology and Applications

• Market platform: Providing 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 developing open-source robotics technology, ethical AI, and decentralized technologies.

• AI agents: Developing AI agents to automate the DAO's operations and governance processes.

Financing Situation

At the end of April 2025, the IDO sold 25 million $DEUS tokens, 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, with specific amounts not disclosed.

(8) GEODNET

https://x.com/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 tokens 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 had its TGE in Q3 2023, currently with a circulating market cap of $57M, total market cap of $179M, and listed on exchanges like Gate and MEXC.

(9) Auki/Posemesh

https://x.com/AukiNetwork

Project Overview

A space computing project based on DePIN, aiming to build 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 providing computing power and storage.

• Data processing: The Posemesh network integrates data to generate high-precision spatial maps and services.

• Data output: Enterprises/developers use the data to optimize business (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, raising approximately 204,700 USDT.

Token Situation

The project had its TGE at the end of August 2024, currently with a circulating market cap of $69M, total market cap of $284M, and listed on MEXC exchange.

(10) peaq

https://x.com/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, gaming, and more.

• Machine economy: By assigning autonomous identities, payment capabilities, and access controls to machines, devices can independently complete 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 liquidity provision in DeFi.

Main 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 connecting devices, sensors, vehicles, and robots, ensuring secure interactions with the network.

• Access control and permission management (peaq Access): A role-based access control mechanism allowing developers to define user groups, roles, and permissions, controlling access to devices or services. For example, in IoT applications, only authorized users can access shared vehicles.

• Decentralized payments and economic models (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 certification, data consistency checks, and community verification.

• Modular DePIN functionalities 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, forming a virtuous 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 multiple institutions including Spartan Group, HV Capital, CMCC Global, and Animoca Brands.

Token Situation

The project had its TGE in November 2024, currently with a circulating market cap of $135M, total market cap of $483M, and listed on exchanges like Gate, Bitget, and MEXC.

(11) CodecFlow

https://x.com/codecopenflow

Project Overview

CodecFlow is an AI operator and robot execution layer based on the Solana blockchain, focusing on solving 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 robot agents).

Product Logic

CodecFlow's product logic 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; modular management of agent states, actions, and security ensures 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 interactions.

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 of 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 through servers without needing high-spec local hardware, enhancing accessibility.

Token Situation

The project had its TGE in May 2025, currently with a circulating market cap of $30M, total market cap of $40M, not yet listed on exchanges, primarily traded on DEX.

CA:69LjZUUzxj3Cb3Fxeo1X4QpYEQTboApkhXTysPpbpump

(12) RICE AI

https://x.com/realRiceAI

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 issues 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, democratizing the robotics industry and accelerating AGI development.

Financing Situation

The project conducted a $750,000 presale of the $RICE token through TokenFi Launchpad in August 2025, accounting for 10% of the total supply, with a project FDV of $7.5 million.

Product Logic

RICE AI's product logic revolves 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 foundry for training robotic foundational models. Developers can subscribe to models for AI application development, receiving discounts for using $RICE. The platform tokenizes data for monetization, addressing high costs, delays, and silos in robot data collection, promoting the shift from simulated data to real-world data, ultimately achieving autonomous learning and collaboration for robots.

Core Features

• Data contribution and rewards: Robot users upload interaction data (such as object grasping, motion trajectories) through VR remote control or automatic collection, earning $RICE.

• Model training and access: The decentralized foundry uses data to train physical AI foundational models, supporting subscription access (with discounts for $RICE payments).

• Governance and data marketplace: $RICE is used for platform governance and data trading, with a portion of platform fees allocated for token buybacks and burns.

• Robot integration: Supports hardware like Minibot M1, integrating GPU processing for real-time learning and collaboration.

• Multi-source data support: Including third-party robots, remote control operations, and self-centered video.

Token Situation

The project had its TGE in August 2025, currently with a circulating market cap of $22M, total market cap of $120M, listed on Binance Alpha, and trading on MEXC.

III. 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 be mainly divided into three categories:

First, new projects that focus on Web3 humanoid robots from the outset, without historical baggage, can integrate on-chain incentive mechanisms and embodied intelligence with a more native logic, possessing strong imaginative potential.

Second, projects originally focused on DePIN, AI, or data that have recently expanded their narratives to enter the robotics sector. 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, although with lower technical barriers, possess strong community operation capabilities and narrative flexibility, and may be the first to produce blockbuster products.

In summary, the current challenges include technological integration, interactive experience, user education, and more. However, 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 with virtual space.

Original link