O-ID, a Techstars-backed startup founded in Tokyo in 2025, builds fully modular humanoid robots aimed at continuous, field-serviceable operation in industrial settings. Rather than building its own factories, the company is leaning on Japan's established automotive and industrial supply chain to reach scale, with a current bill of materials that is already around 85% Japanese by weight.

To dive deeper into O-ID's hardware architecture, supply chain strategy, and deployment roadmap, Humanoid Press interviewed Co-Founder and CTO Simon Gormuzov. Below is the full interview detailing the company's origin story, modular design, manufacturing plans, and path to commercial deployment.

1. Origin

Humanoid Press: What led you and Stian to start O-ID, and how does your background in industrial robotics and his in cyber defense shape the way you've approached the problem?

Simon Gormuzov: When I was working at FANUC and Tokyo Robotics, I experienced two extremes of robotics. On one side, industrial robots are extremely efficient and reliable, but also very rigid. On the other, humanoids are amazing machines, but they are generally too complicated to be practical.

That made me think: why not apply the lessons that make industrial robots successful to humanoids, and build one that is easy to deploy, maintain and mass produce?

When Stian and I met at Waseda, we realized our backgrounds complemented each other well. My experience was mainly in mechanical engineering and industrial robotics, while Stian came from cyber defense and brought a strong systems perspective around resilience, modularity and secure infrastructure. That combination became a big part of how we approached O-ID from the beginning.

Today, most humanoid architectures are still very difficult to mass produce and maintain. Many companies are building their own factories, which can take years, while still depending heavily on globally fragmented supply chains. Maintenance is also a major problem: if a critical part fails or wears out, the entire robot often has to be sent back to the manufacturer, creating significant downtime.

O-ID tackles both problems. Through our deep connections with the Japanese automotive and industrial supply chain, we do not need to build our own factories from scratch. We can work with established manufacturing partners in Japan to produce robots at competitive cost and scale.

Our modular architecture also changes how maintenance works. Instead of shipping the whole robot back, our predictive maintenance system detects irregular performance, and a replacement module can be sent directly to the site. Local staff can swap it in within minutes. That is what keeps the robot practical and economical to operate at scale.

2. Hardware Architecture

Humanoid Press: Can you walk us through the mechanical and electrical design that makes your modules field-replaceable, and how your communication and digital-twin systems keep control accurate?

Simon Gormuzov: We have developed a unique modular architecture, currently under provisional patent, that combines the mechanical, electrical and software interfaces required for seamless module replacement.

The mechanical interface is designed as a rigid, load-bearing connection, while the electrical interface handles power and communication separately. This allows the module to remain structurally part of the robot while still being easy to replace.

From the worker's perspective, the process is simple: they unlock the module, replace it, and the robot takes care of calibration and the more complex steps in the background.

We use high-performance industrial communication standards such as EtherCAT to maintain reliable, deterministic and safe communication across the robot. Our custom software stack allows us to run communication across the system at over 1,000 Hz. Each actuator also has its own digital twin, which models nonlinear characteristics such as joint friction, temperature and other performance factors. This allows us to control the joints more accurately, reduce the sim-to-real gap and build a more sensitive, higher-performance robot.

3. Supply Chain & BOM

Humanoid Press: You've said your bill of materials is already around 85% Japanese by weight. What's the strategic logic behind sourcing so heavily from Japan rather than global suppliers, and how does that support your target of 100,000 robots a year by 2030?

Simon Gormuzov: Our current BOM is already around 85% Japanese by weight, and we have a pipeline to bring the remaining major components into domestic Japanese production, including the sensor, actuator and electronics stacks.

Our strategy is to use the existing and often underutilized capacity of Japanese industrial manufacturers rather than building every part of the manufacturing infrastructure ourselves. This is what allows us to plan for production capacity of 100,000 robots per year by 2030 at a fraction of the capital cost required by companies building their own factories.

At low volume, sourcing globally can sometimes make individual components cheaper. But at humanoid scale, the main cost driver is volume. We believe the larger advantage comes from having tightly integrated suppliers, existing manufacturing capacity and avoiding the capital cost of building an entirely new production system ourselves.

Without giving away too much, at the volumes we are targeting, we expect our robots to be cost-competitive with human labor.

4. Deployment

Humanoid Press: Why start with industrial deployment in Japan specifically, and what does the path from an LOI like the one with Sumitomo Electric look like on the way to real factory floors?

Simon Gormuzov: We plan to start with industrial, human-centered tasks in Japan and move into global exports soon after.

Japan is an ideal starting point because there is already a major labor shortage, particularly in manufacturing. There are more than 200,000 manufacturing establishments in Japan, and the country as a whole is projected to face a labor shortage of around 11 million workers by 2040.

We are targeting work that is currently performed by people but increasingly difficult to staff. That includes machine tending, assembly, inspection and logistics.

The important part is that our robots are designed to work inside existing industrial environments, using existing task profiles. We don't want customers to redesign their factories around the robot. The robot should adapt to the factory and the job that already exists.

With Sumitomo Electric, we have signed an LOI to explore the potential application of their technologies to humanoid robot wire harnesses.

Our next major milestone is unveiling our next-generation, deployment-ready robot at CES 2027. Shortly after that, we plan to begin on-site industrial deployments, with the initial focus on validating uptime, reliability, and task performance in real factory environments. From there, the goal is to move into broader commercial deployments and scale production.

5. Industry Events

Humanoid Press: What should the industry expect to see from O-ID at CES 2027, and is there anything to share around IROS?

Simon Gormuzov: We will be unveiling our next-generation, deployment-ready robot at CES 2027. Once the robot is publicly unveiled, we plan to move shortly afterward into on-site deployments with industrial partners. CES 2027 is confirmed. We do not currently have anything to announce around IROS.

Key Takeaways for Industry Leaders & Developers

• Field-Serviceable by Design: A rigid, load-bearing mechanical interface paired with a separate electrical interface lets local staff swap a module in minutes, guided by predictive maintenance rather than shipping the whole robot back to the manufacturer.
• Capital-Light Manufacturing: Instead of building its own factories, O-ID taps underutilized capacity across Japan's automotive and industrial supply chain, targeting 100,000 units a year by 2030 at a fraction of the capital cost of vertically integrated rivals.
• Deep Japanese Sourcing: An ~85%-by-weight Japanese BOM, with sensors, actuators and electronics moving further in-country, underpins both cost competitiveness at scale and supply chain security.
• Labor-Gap-First Deployment: With Japan facing a projected shortfall of roughly 11 million workers by 2040, O-ID is targeting machine tending, assembly, inspection and logistics roles that already exist — fitting the robot to the factory, not the other way around.

Following the announcement of its FF EAI Robot World 2.0, Faraday Future (NASDAQ: FFAI) has introduced nine new Embodied AI (EAI) device configurations spanning humanoid, quadruped, and wheel-legged forms. Highlighting the launch is the FF Master Mini — North America's first compact humanoid priced under $10,000 — and the flagship FF All-New Futurist, powered by NVIDIA Jetson Thor delivering up to 700 TOPS of onboard compute.

To dive deeper into the commercialization, hardware cost architecture, and AI training methodology behind this rollout, Humanoid Press interviewed Jerry Wang, Global Executive Chairman of Faraday Future. Below is the full interview detailing FF's hardware manufacturing strategy, multi-form data flywheel, edge computing roadmap, and deployment ecosystem.

1. Market Positioning & Cost Architecture

Humanoid Press: Sub-$10,000 is a major price milestone for competition and educational humanoids. What key hardware or supply chain choices allowed Faraday Future to reach the Master Mini's $9,990 price point without compromising on degrees of freedom or compute capability?

Jerry Wang: FF is leveraging mature, standardized robotics hardware and established supply-chain capabilities to keep costs down and our technology core intact. With the Master Mini, our goal was to preserve the capabilities that matter for education, development, and competition, while leveraging mature, standardized robotics hardware and an efficient supply-chain model.

The Master Mini maintains 22 degrees of freedom (DoF) and provides an open development environment that supports ROS 2 SDKs alongside simulation and training tools, giving students, developers, and researchers a capable platform at a much more accessible price point.

An important part of FF's broader robotics strategy is what we call our "Light on Four, Heavy on Four" model — meaning we aim to remain light on capital, assets, marketing, and sales while investing our resources heavily in product strength, AI, real-world utility, and user experience.

That allows us to take advantage of established hardware platforms and supply-chain capabilities rather than vertically integrating every component ourselves, and then focus FF's differentiation increasingly at the AI, software, application, and ecosystem layers.

2. The EAI Ecosystem & Multi-Form Data Flywheel

Humanoid Press: YT Jia mentioned that multi-robot form factors generate diverse scenario data to train a central "EAI Brain." How does collecting data across quadrupeds, wheel-legged, and humanoid platforms accelerate model iteration faster than training on a single form factor alone?

Jerry Wang: The core advantage of a multi-form strategy is that each robot form encounters the physical world differently, so together they can generate a much richer and more diverse set of real-world scenarios than any single form factor could generate alone.

A humanoid, a quadruped, and a wheel-legged robot have fundamentally different mobility, perception, interaction, and task requirements. That diversity exposes the EAI Brain to a broader range of environments and physical use cases, helping FF develop intelligence and Skills that can be adapted across platforms rather than optimized around only one type of robot.

3. Edge Compute & Vision-Language-Action (VLA) Models

Humanoid Press: With the Futurist Ultra leveraging NVIDIA Jetson Thor for 700 TOPS of compute alongside Vision-Language-Action (VLA) models, what specific enterprise or industrial tasks do you expect this level of onboard compute to solve first?

Jerry Wang: We believe the significance of 700 TOPS of onboard computing power is not simply about making the robot move faster. It is about enabling the Futurist Ultra to understand more complex environments, reason about what it sees, and move from executing individual commands toward completing entire tasks.

In the near term, we believe some of the most practical enterprise opportunities will be structured, repeatable workflows where perception, reasoning, and physical interaction have to work together. That could include:

• Inspection-related activities
• Navigating enterprise or industrial environments
• Recognizing equipment and objects autonomously
• Gathering and interpreting multimodal information
• Performing complex sequences of actions rather than requiring a separate instruction for every step

Ultimately, we see this expanding into more complex Physical AI applications as the EAI Brain, Skills, and real-world data ecosystem mature.

4. Commercialization & Overcoming Adoption Barriers

Humanoid Press: Hardware cost is only part of the equation — deployment friction is often the bigger hurdle. How do the "6+1" direct sales network and RoboShare platform make it easier for universities, security firms, and industrial facilities to onboard these machines quickly?

Jerry Wang: We believe the real challenge in enterprise robotics is not just lowering the hardware cost. It is lowering the total barrier to deployment and giving customers a practical path from evaluation to real-world operation. That is exactly what our "6+1" model and RoboShare are designed to address.

On the "6" side, FF has built six distinct routes to market:

• B2B Direct
• FF Par
• E-commerce
• Marcom & Video Sales
• Community Sales
• Mcube

The idea is to give different customer types multiple ways to discover, evaluate, purchase, and deploy our EAI products instead of relying on a traditional, single-channel sales model. All nine EAI Device configurations announced at our 919 event are already on sale and available for delivery, which is critical when customers want to move rapidly from evaluation to deployment.

But we believe the bigger opportunity is the "+1" — RoboShare. RoboShare provides rental and asset-onboarding options, which allow a university, security operator, or industrial customer to evaluate a robot in its actual operating environment before committing to a larger capital purchase or fleet deployment. RoboShare has also signed an MOU with Hifivebot, a U.S. robot rental platform, as part of developing that model.

Just as importantly, we are not asking customers to start with a robot and figure out the application themselves. We introduced four Industry Productivity Solutions covering K-12 Education, Research, Security, and Inspection, specifically aimed at reducing deployment barriers and creating replicable paths to practical robot applications. That shifts the conversation from "What can this robot theoretically do?" to "How can I put it to work in my organization today?"

Key Takeaways for Industry Leaders & Developers

• Price Disruptor: The $9,990 Master Mini provides a complete ROS 2 development environment, opening humanoid research and soccer robotics competition to high schools and undergraduate labs previously priced out.
• Capital-Efficient Manufacturing: FF's "Light on Four, Heavy on Four" strategy avoids heavy vertical component manufacturing, choosing instead to aggregate existing supply chains and focus value capture on software, VLA foundation models, and system integration.
• Multi-Form Data Collection: Rather than betting solely on humanoids, FF's inclusion of quadruped (Aegis Hyper) and wheel-legged (Aegis Mega) robots feeds environmental navigation data back into a unified EAI Brain.
• Flexible Onboarding: Through RoboShare, enterprises can pilot platforms without initial CapEx commitments.