At the start of this year, when Elon Musk returned to Davos, he once again delivered his provocative forecast: in the future, robots will outnumber humans on Earth.
It's clear that on a global scale, AI and robotics have become the two dominant technology topics: artificial general intelligence (AGI), now nearing a breakthrough, and robotics, which is leaving the confines of research labs and moving toward large-scale automation of human physical labor. This year, in addition to AI, embodied intelligence has also emerged as a key sector within the cryptocurrency industry. Below are several Robotic sector projects worth watching.
OpenMind
On August 4, 2025 (UTC), Silicon Valley-based intelligent machine infrastructure company OpenMind announced it had secured $20 million in financing. The round was led by Pantera Capital, with participation from Ribbit, Sequoia China, Coinbase Ventures, DCG, Lightspeed Faction, Anagram, Pi Network Ventures, Topology, Primitive Ventures, Amber Group, and several prominent angel investors.
OpenMind develops open-source software that enables robots to think, learn, and work. Its native open-source AI robot operating system, OM1, allows for the configuration and deployment of AI Agents both in digital and physical environments. Users can create AI agents that run in the cloud or deploy them on physical robots in the real world.
Simply put, OpenMind’s OM1 is building the “AI brain” for robots. This brain can coordinate multiple AI Agents, interact with a variety of LLMs, and aggregate data from diverse sources (such as posting content to social media on behalf of a user). Because OM1 is open source, it’s a highly adaptable robot operating system—much like how Android is hardware-agnostic in the smartphone world.
OpenMind has also developed a blockchain-based robot identity network called FABRIC, designed to establish a verifiable trust layer shared between humans and robots. Humans can earn badges by sharing location data, evaluating robot behavior, and building applications. For robots, any device running the OM1 system joins the FABRIC network, gaining a unique, verifiable identity and enabling on-chain tracking of commands, operation logs, ownership, and related activity.
In December 2025 (UTC), OpenMind and stablecoin issuer Circle jointly announced the launch of an autonomous robot payment system based on the x402 protocol. As robots become more advanced, they will no longer be just tools for executing tasks—they'll become autonomous economic agents, purchasing compute, data, and skills, and even hiring other robots or humans to accomplish complex goals.
CodecFlow
CodecFlow delivers a unified platform that operates seamlessly across the cloud, edge, desktop, and robot hardware, supporting both modern APIs and legacy systems. The platform standardizes diverse robot sensor inputs into a universal format and modularizes complex robot actions, so developers and users don’t have to design robots from scratch. With CodecFlow, perception, decision-making, and control are networked and interoperable, rather than fragmented or tied to specific hardware.
AI-powered operators leverage perception and real-time reasoning to react to UI changes in software or to environmental changes in the robot’s surroundings. This approach solves the traditional fragility found in pre-scripted robot automation, which often fails with even minor variations. In essence, CodecFlow captures screen images, camera feeds, or sensor data, uses AI to process these external inputs for observations and instructions, and then executes decisions through UI interactions.
Peaq
On March 27, 2025 (UTC), DePIN Layer1 protocol Peaq raised $15 million in a round led by Generative Ventures and Borderless Capital, joined by Spartan Group, HV Capital, CMCC Global, Animoca Brands, Moonrock Capital, Fundamental Labs, TRGC, DWF Labs, Crit Ventures, Cogitent Ventures, NGC Ventures, Agnostic Fund, and Altana Wealth.
While Peaq initially positioned itself as a DePIN platform, it also launched a Robotics SDK in September last year, enabling robots to gain autonomous identities, process payments, verify data, and access on-chain network economies. Now, any ROS2-compatible robot can join the Peaq network economy and use its standards to transact with humans and other robots.
Last year, Peaq launched the “RoboFarm” robot RWA project on DualMint, building a robot-operated farm in Hong Kong that automates 80% of its agricultural production. The farm’s lettuce, spinach, and kale are sold locally in Hong Kong, and NFT holders receive an estimated 18% annualized yield.
Axis Robotics
Axis Robotics is building distributed, scalable infrastructure for Embodied Intelligence (Physical AI). They believe a “Simulation First” approach is the best way to overcome the bottlenecks of data scarcity and model generalization in robotics. By combining low-cost, large-scale data collection with proprietary data augmentation, they’ve achieved major advances in data quality, richness, and scale. Every data asset is also tracked with trustworthy on-chain provenance, establishing a core data engine to fuel general robot intelligence (RGI).
Axis has revolutionized how robot training data is sourced. Competing projects typically crowdsource user videos of specific actions performed in the real world via smartphones or smart glasses, enabling broad, low-barrier participation. However, this approach produces data lacking in physical realism, depth, and 3D continuity.
Axis solves this by using simulation, generating vast numbers of diverse virtual scenarios (lighting, angles, friction, dynamics, etc.) that train models to perform under challenging conditions, thus improving generalization. Their hybrid strategy combines scarce real-world data with massive synthetic data, using GPU-accelerated meta-data augmentation to vary lighting, textures, and physical properties within each scene. These virtual environments are flexible, not hard-coded, enabling the creation of countless scenarios for robots to tackle. The cost per scenario is low, output is high, and this data-driven approach to model optimization is validated by leaders like Google and NVIDIA.
Axis launched its first community-accessible simulation learning project, “Little Prince’s Rose.” In this project, users remotely operated a robot in a browser-based simulation to water a plant. Analysis of user actions enabled the robot to learn the task. This approach maintained the low entry barrier of video uploads while building a native 3D-aware VLA (Vision-Language-Action) foundation model to enhance spatial reasoning—something video data alone can’t provide.
In just 5 days, thousands of users with no robotics background contributed tens of thousands of high-quality, training-ready trajectories. Axis used this data to train a policy model and successfully deployed it on a real-world Franka robotic arm, completing the end-to-end loop from task generation, community data collection, and augmentation to model training and real-world deployment.
With this approach, one hour of real data can translate to 1,000 hours of training data, dramatically reducing the cost of robot model generalization.
During the Lunar New Year beta, again in just 5 days, 18,000 non-expert users completed 27 new tasks, contributing over 100,000 data trajectories. The test validated high task randomization and compatibility with various robot types, including wheeled and dual-arm robots.
Axis will officially launch its core product in late March, with plans to open-source the world’s largest Franka-arm-based simulation dataset by late April or early May—fully meeting strategy and model training needs. As a Crypto-AI robotics project, Axis is also advancing industry adoption: collaborating with an automaker to automate production; partnering with a soon-to-IPO compute firm on virtual assets and world models; and forging deep partnerships with embodied entity firms for simulation data and model training. These steps reveal the unique externalities of Crypto projects.
GEODNET
GEODNET is a decentralized network delivering centimeter-level, real-time dynamic positioning for drones, robots, and other devices. It operates over 21,000 active base stations across 150+ countries. Over the past year, the project generated $7 million+ in revenue, with growth each quarter.
While often categorized as DePIN, the rise of real-world robotics is expected to drive broader demand for high-precision real-time positioning data. In February 2025 (UTC), Multicoin announced the acquisition of $8 million worth of $GEDO tokens from the GEODNET Foundation.
BitRobot
BitRobot Network, co-developed by FrodoBots Lab and Protocol Labs, enables distributed robotic work and collaboration. Key components include: Verifiable Robot Work (VRW, a network reward metric for defining and validating robot tasks); Equipment Node Tokens (ENT, unique robot identifiers as NFTs); and subnets, which are resource clusters that execute tasks and create value for the network.
On February 14, 2025 (UTC), FrodoBots Lab announced a $6 million seed round, bringing total funding to $8 million.
FrodoBots Lab also sells robots: Earth Rovers, which resemble real-world Mario Karts and are priced at $249, can be remotely piloted by players in the global treasure hunt game ET Fugi via browser. The data generated supports researchers testing the latest AI navigation models. ET Fugi is BitRobot’s first subnet.
A future robot, Octo Arms, will let users remotely control robotic arms to solve 3D puzzles and compete in games.
The “subnet” concept is abstract—any cluster contributing to the network ecosystem (or a specific project or event) is a subnet, such as ET Fugi or SeeSaw by Virtuals.
SeeSaw
BitRobot’s fifth subnet, SeeSaw, is a robot training data sharing app launched by Virtuals in October last year. In SeeSaw, users upload videos of their daily activities to earn rewards. These videos, capturing everyday actions like tying shoelaces or folding clothes, come from a global user base and are used to train robots.
Auki
Auki’s decentralized machine perception network, Posemesh, connects humans, devices, and AI. The core DePIN (Decentralized Physical Infrastructure Network) architecture lets robots, AR glasses, and other devices share real-time location and sensor data, building a collaborative spatial understanding of the physical world for robots, AR, and AI.
Posemesh defines multiple node types: compute nodes provide processing power; motion nodes (robot endpoints) upload location and sensor data; reconstruction nodes generate 3D map models; and domain nodes manage 3D space. Each node earns $AUKI tokens based on its contribution, powering a self-evolving machine vision network.
The network prioritizes privacy, preventing any single entity from monitoring user spaces, and is applicable to retail (product placement optimization), property management (asset tracking), event navigation, and construction or renovation.
Their Cactus AI spatial computing platform has begun pilots with Toyota Material Handling and Sweden’s Stora Coop supermarket.
XMAQUINA
XMAQUINA is a DAO that enables retail investors to participate in robotics company investments. The DAO raised $10 million through phased sales of its $DEUS token, using the proceeds to acquire equity stakes in six robotics firms: Apptronik, Figure AI, Agility Robotics, 1X Tech, NEURA Robotics, and Robotico. Some investments have already realized profits, with single-deal returns exceeding 100%.
PrismaX
On June 17, 2025 (UTC), PrismaX announced an $11 million funding round with backers including a16z CSX, Volt Capital, Blockchain Builders Fund, Stanford Blockchain Accelerator, and Virtuals.
PrismaX is building an open coordination layer connecting remote operators, robot users, and robotics companies. Operators can connect with users, remotely control robots for real-world tasks, and collect valuable data. They can also request logistical, advertising, or other real-world services.
PrismaX’s remote operation protocol allows companies to find skilled robot operators for complex tasks. Operators can stake network tokens to boost trust and increase their chances of winning high-value assignments. Staking rewards depend on both the amount staked and work quality, with increased efficiency yielding extra bonuses.
Data from remote operations is used to train robots, improving their autonomy, which in turn boosts operator efficiency and accelerates the transition to highly or fully autonomous machines.
NRN Agents
NRN grew out of AI Arena, a real-time training game for AI Agents. On October 28, 2021 (UTC), developer ArenaX Labs raised $5 million in a seed round led by Paradigm Capital and joined by Framework Venture Partners. On January 9, 2024 (UTC), ArenaX Labs closed a new $6 million round led by Framework Ventures, with support from SevenX Ventures, FunPlus/Xterio, and Moore Strategic Ventures.
While the core model is still data collection and reinforcement learning for robots, NRN uses its gaming expertise to turn robot data collection into a browser-based game. Users intuitively control simulated robots, generating behavioral data in gameplay that is then used to train real-world robotic systems.
Currently, the project is focused on robotic arms (RME-1) to validate data collection, real-time learning, and adaptability.
Notice:
- This article is republished from [BlockBeats] and is copyrighted by the original author [Cookie]. For concerns about this republication, contact the Gate Learn team, who will address it promptly according to established procedures.
- Disclaimer: The views and opinions expressed in this article are those of the author alone and do not constitute investment advice.
- Other language versions are translated by the Gate Learn team. Unless Gate is expressly credited, translated articles may not be reproduced, distributed, or plagiarized.
