Genisom AI ships over 10,000 robots in under three years

Genisom AI has shipped more than 10,000 robots in under three years, a milestone the company touts as proof that deployable robotics at scale is finally breaking out of the lab. The claim anchors a debut at ICRA 2026 that the Beijing-based startup frames not as a concept but as a real-world, manufacturer-ready platform business. At ICRA in Vienna, Genisom AI showcased the M1, its industrial-grade quadruped, alongside the L1-series, signaling a pivot from demonstrations to factory floors.

Testing shows the GENISOM M1 is engineered for payload-driven work. It carries a continuous walking payload rated at 30 kilograms (66.1 pounds), with a payload-to-weight ratio approaching 1 to 1 and up to five hours of runtime depending on payload and operating conditions. The platform carries an IP67 protection rating, underscoring the company’s emphasis on field durability rather than pristine lab benches. The numbers are not just marketing: the M1’s performance is underpinned by an in-house joint actuator module, the P85MAX-S, which delivers up to 180 newton-meters of peak torque within a compact 86-millimeter diameter package weighing roughly one kilogram.

The decision to develop actuators in-house is not incidental. Documentation indicates that Genisom AI seeks tight hardware-software integration from the ground up, arguing that direct control of locomotion performance, payload capacity, and system reliability comes when the company owns the motor, transmission, and control stack in one loop. The company reports that this approach helps it tailor motion profiles to real-world tasks rather than rely on off-the-shelf components that can force compromises in rugged deployment scenarios.

Genisom AI is quick to place its work in the context of a broader industry push toward scalable robotics platforms. Like Unitree, the company says it is building around manufacturable platforms designed to move beyond laboratory demonstrations and reach customers at scale. The emphasis on real-world deployment, paired with in-house core technologies and a software stack intended to generalize across use cases, distinguishes Genisom AI from many theory-first or early-stage players still chasing pilot programs.

Beijing-based and moving with a production mindset, Genisom AI frames M1 as a modular, enterprise-ready system rather than a one-off research rover. The company has repeatedly framed the 10,000-unit milestone as evidence that a true hardware-software fusion can scale in robotics, not just in software or in narrowly scoped demos. The payoff, executives suggest, lies in customers across industries who want reliable, repeatable motion and predictable serviceability on a shop floor or in remote environments.

Industry observers will be watching how the M1 and its platform family perform in sustained field use, and whether a higher-rate supply chain for core modules such as the 180 N·m P85MAX-S holds up under serial production. The five-hour runtime figure, while compelling, hinges on payload and operating conditions and will be a key constraint for operators planning continuous-duty cycles. Thermal management, battery cadence, and field maintenance will shape the next wave of deployments, particularly as Genisom AI pushes beyond prototyping to long-duration customer service contracts and cross-industry integrations.

In short, Genisom AI is not just claiming a milestone; it is tying it to a pragmatic path toward scalable, field-ready robotics. If the 10,000-unit mark translates into durable deployments and repeatable performance on customer sites, the industry may have a tangible blueprint for turning laboratory motion into reliable, repeatable automation.