In 2025, humanoid robots are moving from science fiction into reality. From Tesla's Optimus to Figure AI's Figure 01, the capabilities of general-purpose humanoid robots are rapidly expanding with the support of large language models. According to Goldman Sachs, the humanoid robot market could reach $154 billion by 2035. A massive trillion-dollar market is now attracting the world's top technology companies and brightest minds to join the effort.

However, as the "limbs" of robots become increasingly advanced, a more fundamental question arises before us: how can we build a sufficiently intelligent, open, and secure "brain"? As millions of robots enter homes, hospitals, and cities, how will they collaborate, exchange value, and seamlessly integrate into human society?

Jan Liphardt, a Stanford University professor and founder of OpenMind, provided his insights. After securing $20 million in funding led by Pantera Capital in August 2025, OpenMind accelerated its development, launching a series of products ranging from the underlying operating system to upper-layer payment protocols, gradually outlining a complete blueprint for its "robot brain."

Jan Liphardt, Founder of OpenMind

OpenMind's core business is providing cloud-based cognitive services to enterprises through a SaaS model. However, they have keenly recognized that as robots become independent economic participants, blockchain will play a crucial role in areas such as payment systems, identity verification, data privacy, and collaborative governance.

Recently, OpenMind's collaboration with stablecoin issuer Circle, as well as the deployment of robot charging stations on the streets of San Francisco, mark the initial realization of this vision. Robots can now independently pay for charging using USDC, which may signal the dawn of the "Machine Economy" era.

At the same time, OpenMind is also building a dedicated app store for robots, allowing users to download applications and skills to their robots in one place, just as they would customize their mobile apps on the Apple App Store or Google Play Store. The app was launched in the OpenMind App Store last week.

In this exclusive interview, we had an in-depth conversation with the founder of OpenMind about the philosophy behind building "brains" for robots, the design principles of the modular operating system OM1, and how to build a future of efficient collaboration between machines and between machines and humans through the FABRIC protocol and blockchain technology. He shared OpenMind's technical roadmap and provided profound insights on key issues such as the developer ecosystem, remote operation, and data privacy.

The following is the content of the exclusive interview:

Set up a "bank account" for the robot

In December 2025, OpenMind, in collaboration with stablecoin issuer Circle, announced the launch of a robot autonomous payment system based on the x402 protocol. As robotic capabilities advance, robots will no longer merely serve as tools for executing tasks but will begin to assume roles within autonomous economies. They will need to purchase computing power, data, and skills, and even hire other robots or humans to accomplish complex tasks.

To achieve this, a financial system specifically designed for machines, without requiring human intervention, has become essential. The traditional banking system is clearly unprepared for this, while cryptocurrencies and blockchain technology, with their native digital and decentralized characteristics, have become the most natural choice.

BlockBeats: What were you doing before founding OpenMind? What prompted you to pursue this venture?

Jan: I am a professor of engineering at Stanford University, but I am currently fully committed to OpenMind. I founded this company because I believe the traditional robot software stack is not well-suited for complex and dynamic environments like hospitals and homes.

OpenMind is an American technology company, but its core is not a cryptocurrency business; rather, it is an enterprise-level SaaS cloud cognition company. Our business model is similar to other enterprise SaaS companies, primarily generating revenue by establishing standard cloud-based APIs.

As for blockchain, it has some interesting characteristics in tracking information and building financial systems. Looking ahead, we foresee autonomous machines interacting with other machines and even humans to collaboratively accomplish tasks. Blockchain offers a set of potential technical solutions here, especially in areas such as machine payment systems, identity management, collaboration, and governance.

BlockBeats: Recently, OpenMind announced a collaboration with Circle on the x402 protocol. Can you introduce how this collaboration came about? Why is it so important?

Jan: In fact, as early as May last year when Coinbase Developer Platform first released x402, our robot was already one of the first partners to support x402. In our software, we directly integrated the payment system into the robot's "brain," with the goal of enabling the robot to interact with external infrastructure.

We have always been thinking: what would a payment system look like if it were designed not around humans, but around machines? This question ultimately led to our collaboration with Circle. The core idea is that machines don't have pockets, fingerprints, eyes, or passports, but they are exceptionally good at writing code and using APIs.

Media coverage of OpenMind's collaboration with Circle

Therefore, from our perspective, for a robot, purchasing goods and services through a digital payment system often feels more natural than using credit cards or cash. What we are building with Circle is a location-based payment system. When two devices come close to each other, they can directly exchange money.

A practical example is the charging stations we set up on sidewalks in San Francisco for autonomous machines. When a robot approaches, the system detects its presence, activates the charger, and the robot can purchase electricity using the stablecoin USDC.

BlockBeats: Why do you think it is crucial for robots to have this autonomous purchasing capability?

Jan: Take autonomous taxis (robotaxis) as an example. They definitely require a solid payment infrastructure. Of course, they could use fiat currency, but that feels cumbersome; using credit cards is an option, but it feels outdated. NFC-based protocols are somewhat more interesting, but when we repeatedly "communicate" with highly advanced robots, we keep hearing that they are more than willing to use cryptocurrency as a payment tool.

These machines are inherently good at handling digital infrastructure, and in practice, cryptocurrencies could be extremely convenient for autonomous machines to make payments.

If a humanoid robot walks into a bank, the bank would call the police. Human-centered banks have no real conceptual model for an autonomous physical entity capable of managing funds and making independent decisions. Traditional banks ask for your name, social security number, passport, address, and birthplace—questions that hold no meaning for an autonomous humanoid robot.

Institutions like Bank of America currently do not have the concept of providing bank accounts or credit cards to non-biological thinking machines. Perhaps this will change in the future, and banks may expand their services to non-biological clients. But today, if you are a smart machine, the only viable option is cryptocurrency.

BlockBeats: So this is more of an advantage rather than a strict requirement. A payment system between bots doesn't necessarily have to use cryptocurrency, but it would be a more elegant solution, right?

OpenMind: If a humanoid robot walks into a bank, the bank would call the police. Human-centered banks have no real conceptual model for an autonomous physical entity capable of managing funds and making independent decisions.

Traditional banks will ask for your name, social security number, passport, address, birthplace, and other such information, which is meaningless for an autonomous humanoid robot.

BlockBeats: What is the cost to deploy such a charging station?

OpenMind: The hardware costs are approximately three hundred dollars. As for electricity costs, they depend on the operator and are not determined by us. What we build is the software and infrastructure.

But this is just a small example. More broadly, as machines become aware and more intelligent, they will want to buy and sell many different things: real-time data, new models and skills, computing power and storage. They may take on jobs and tasks and collaborate closely with humans.

All of this requires solid infrastructure to coordinate payments and collaboration between machines and humans. We are not a charging station company. We are striving to provide the full set of capabilities that intelligent machines need, enabling them to be safe and useful for people everywhere.

OM1 and FABRIC: From "Individual Intelligence" to "Collective Collaboration"

To truly integrate robots into society, a powerful "brain" is first needed to understand the world—an advanced operating system. OpenMind's OM1 aims to provide a single robot with unprecedented environmental perception, language interaction, and spatial reasoning capabilities through a modular multi-model architecture.

However, true intelligence emerges from collaboration. The vision of the FABRIC protocol is even grander: it aims to become the "TCP/IP" of the robot world, enabling machines of different brands and forms to communicate and collaborate freely, just like humans, and collectively form an intelligent physical network.

Robots Equipped with OpenMind OM1 Witness the Launch of the First Humanoid Robot ETF, KraneShares KOID

BlockBeats: For readers who are unfamiliar, could you explain OM1 operating system and FABRIC protocol? Let's start with OM1.

Jan: OM1 is a modular operating system designed for human-oriented robots. It is not suitable for industrial robots, but rather for those robots that interact with people and children, live in your home, or play roles in hospitals and schools.

These robots need to understand their spatial environments, be multilingual, comprehend the structure of buildings, and perform reasoning in space. Traditional robot operating systems (ROS) do not actually provide these capabilities.

OM1 is designed to be modular, much like LEGO blocks that can be snapped together. In practice, we run about 5 to 15 models in parallel, each responsible for different capabilities, such as vision, audition, speech generation, and fusing data from multiple sensors into a continuous view of the environment, including people, pets, rooms, and other aspects of the surroundings.

A quadruped robot equipped with OpenMind developer tools

FABRIC, on the other hand, is still in a very early stage, not yet built, and will require a long time to develop. Moreover, we will only be one of many contributors. If OM1 is about making a single machine intelligent, then FABRIC is about enabling multiple machines to work together, whether with other machines or with humans.

BlockBeats: What was the original intention behind building the FABRIC protocol?

Jan: The initial trigger came from a moment in the real world. One of our humanoid robots was crossing the street, and we saw a Waymo (autonomous vehicle) approaching. Waymo is a robotic car, and we were curious about what would happen at the crosswalk.

The result was smooth. Waymo came to a stop. It might have recognized the humanoid robot as a human, waited for it to cross the street, and then continued driving.

This makes us wonder—if Waymo knows about the existence of the humanoid robot, and the humanoid robot also knows about another robot—the autonomous taxi—wouldn't that be useful?

This prompted us to start thinking about a system that would allow one machine to communicate with another entirely different machine—regardless of manufacturer or form, whether it has wheels, arms, or legs. We are looking for something like a "smartphone" or "Zoom" for machines, a way for physically nearby machines to collaborate effectively.

BlockBeats: You said it will take a long time to build FABRIC. Why is that?

OpenMind: There are many reasons. Machines come in various forms, such as wheels, legs, and claws. There are also numerous manufacturers. Additionally, the types of data that machines want to share are diverse. Moreover, there are regional-specific requirements, including different languages, capabilities, and use cases.

You can relatively quickly build general infrastructure at a basic level, but to create everything that is needed, a great deal of work must be done by many people with different skills in various locations.

BlockBeats: When an AI product runs multiple models, token costs can become very expensive. Will this become a cost issue for OM1's users and developers?

OpenMind: Cost is always a concern, but there are many ways to address it. Some of the models we run are open source, and many of today's top-performing models are also open source. Therefore, the cost essentially comes down to computation and electricity. Some of our models are very small and simple. For example, there are models focused on safety, ensuring that humanoid or quadrupedal robots won't trip over shoes, carpets, or stairs.

In general, we can run most of the stack on a single NVIDIA A4 or a Mac M4/M5 level chip. In terms of cost, this is roughly equivalent to running something on your own laptop. We don't see cost as a major barrier.

Developer Ecosystem: How Does BrainPack Break Through the Challenges of Robot Development?

In the era of software-defined hardware, a thriving ecosystem is key to the popularization of technology. Just as the success of the iPhone was inseparable from its vast App Store developer community, humanoid robots face challenges that hinder developers from fully showcasing their talents. These challenges include high hardware costs, fragmented development systems, and the lack of intelligent systems.

Meanwhile, OpenMind is building a series of robot software ecosystems aimed at breaking through this bottleneck, including the intelligent operating system OM1, the collaborative network FABRIC, and the robot's "pluggable brain," BrainPack. In addition, OpenMind has just launched its first robot app store, allowing users to download applications and skills to their robots in one place, just as they would customize mobile apps on the Apple App Store or Google Play.

BlockBeats: In your opinion, what is the current state of the robot developer ecosystem? What might be the biggest obstacles?

Jan: Almost everyone is enthusiastic about powerful and safe humanoid robots, from students in robotics classes to senior developers at companies like Meta or Google. The issue isn't a lack of enthusiasm, but rather two key factors. First, there is a very limited number of advanced humanoid robots in practical use. Second, almost all current robots use custom, poorly documented methods to access data, internal states, and control their own behaviors.

Currently, there is a near-complete lack of general-purpose systems for adding and improving advanced capabilities in humanoid robots. Many fundamental issues, such as battery management and navigation, can be addressed with existing software like ROS2. However, enabling robots to understand their spatial environment, entertain people, learn new skills, and perform well in highly dynamic environments such as homes, hospitals, and schools, currently has almost no available solutions.

OpenMind aims to help bridge this gap by developing open-source software for social robots, enabling developers around the world to easily understand, learn, and contribute to this rapidly evolving field.

BlockBeats: You describe BrainPack as a small step toward the "iPhone moment" for humanoid robots. What does BrainPack specifically bring?

Jan: One of the main issues today is that there is a huge variation among different humanoid robots. For software developers, it can take a long time just to learn the specific details of one robot before they can write something useful.

BrainPack is designed to solve this problem. You can imagine it as a backpack containing a computer that can connect to a robot. If your software runs on BrainPack, it abstracts away the hardware differences between various robots. This means developers can focus on functionality without worrying about the unique APIs or SDKs of each robot.

BrainPack Installed on the Robot

If the software runs well on BrainPack, it is likely to run on a variety of robots, regardless of whether they have two legs, four legs, wheels, or are tall or short. BrainPack also comes with a set of standardized sensors, so developers don't need to deal with different sensor formats or data protocols. In addition, BrainPack connects directly to our cloud infrastructure, making it easy to leverage remote computing.

BlockBeats: Besides charging stations, what other infrastructure might OpenMind deploy in the future to demonstrate the capabilities of OM1 and the FABRIC protocol?

OpenMind: Another example is the work we have already started with NEAR AI. This project uses NVIDIA H100 and H200 GPUs to enable confidential computing.

Confidential computing means that robots can run models anywhere on Earth while ensuring the data being transmitted back and forth remains confidential. As a result, a robot in San Francisco can have its "brain" hosted thousands of miles away. This also means that individuals with the appropriate hardware (H100 and H200) can provide confidential computing nodes for AI and robotics.

Trust, Privacy, and the New Economic Model

The practical application of technology ultimately returns to society. In addition to technical challenges, the widespread adoption of robots also faces a series of socio-structural issues, including trust, safety, regulation, privacy, and public acceptance. OpenMind believes that open source is the cornerstone of building trust, allowing people to "see" how a robot's "brain" works. At the same time, collaborating with projects like NEAR and using confidential computing technologies to protect data privacy will be key to gaining public trust. A future deeply involving robots will inevitably give rise to entirely new job roles and economic organizational models.

BlockBeats: You mentioned on X that teleoperation could become a real career category in the future. Could you explain this idea in more detail for our readers?

Jan: From a very practical perspective, today's robots still require a lot of assistance. They sometimes get stuck, sometimes don't know the correct answers, and sometimes make mistakes.

In these situations, having a human nearby the robot, either physically or through close monitoring, is extremely useful. Another aspect is trust. Many people are not yet comfortable with robots making fully autonomous decisions, so having a "human in the loop" can help people feel more at ease.

In addition, remote operation has created new opportunities. You no longer need to be in a specific location to perform certain types of work. Based on your skills, you can assist in operating or supervising a robot that is located thousands of miles away, even on a different continent. This opens up a wide range of new economic and career opportunities.

BlockBeats: What plans does OpenMind have to help regions or societies better accept humanoid robots?

Jan: Trust is the foundation. If people feel afraid, the adoption rate will be very slow. That's why our core software is open source. We want people to be able to see inside the robot's "brain" and understand how it works.

Another unresolved issue is ownership. Will robots be purchased by employers? Or by individuals for home use? Or will they be shared by communities? A model similar to shared car ownership might emerge, in which a group purchases a robot and shares the benefits from the work it performs.

We don't yet know which model will dominate, but there is ample room for many new ways to organize work and create value around robots.

BlockBeats: Let's get back to the issue of privacy. You mentioned the collaboration with NEAR. Could you explain more clearly why this collaboration with NEAR is important?

Jan: The core technology here is confidential computing, which is directly integrated into the NVIDIA H100 and H200 GPUs. In principle, anyone who owns these GPUs can connect them to the internet and provide secure computing services to others.

NEAR is precisely very fast, highly capable, and deeply interested in building the infrastructure needed to make this kind of access practical and scalable. That's what facilitated the collaboration. But at a fundamental level, confidential computing is a capability available on every H100 and H200 GPU.

BlockBeats: How large is the OpenMind team now?

OpenMind: We currently have about twenty people, located in San Francisco and Hong Kong.

BlockBeats: What do you expect to be the main product or revenue driver for OpenMind in the next three years?

OpenMind: Our fastest-growing revenue comes from enterprise AI, particularly through cloud-based model delivery and robot-centric computing services. Customers pay directly for these services. Another significant area is revenue sharing with robotics companies. We collaborate with them to co-develop products, which are then sold in regions such as Europe, the Middle East, and the United States.

BlockBeats: Many people are concerned about the scale of capital expenditures in today's AI field. Do you think OpenMind needs a large amount of funding to continue its development, or can it achieve self-sustainability relatively quickly?

OpenMind: This is a bigger issue, but we have a different view on the idea that it would take tens of billions of dollars to build useful models.

We have already seen some strong examples, such as DeepSeek, whose development budget was much smaller than models like ChatGPT. From our experience, many of the models we need can be built with significantly less capital than people typically assume.

Therefore, we are cautiously optimistic that making meaningful progress in robotics or AI does not necessarily require computational resources in the tens of billions or even hundreds of billions of dollars.

BlockBeats: Finally, is there anything you would like to say to the developer or user community in China?

OpenMind: This is an extremely rare moment. A brand new technology is emerging that enables machines to perform tasks that were previously only possible for humans. This will have a profound impact on education, healthcare, manufacturing, and many other areas of life.

For software developers, the opportunity is no longer just about building applications for phones, but about building applications for thinking machines. It's still early days, but progress is happening very quickly. I strongly encourage developers to learn about robot operating systems, humanoid robot platforms, and how to build applications for them, in order to be well-prepared for the significant advancements that are coming.

Click to learn about BlockBeats' job openings.

Welcome to join the official Lulin BlockBeats community:

Telegram Subscription Group:https://t.me/theblockbeats

Telegram discussion group:https://t.me/BlockBeats_App

Official Twitter account:https://twitter.com/BlockBeatsAsia