How they use Nebius AI Cloud

RoboForce uses Nebius AI Cloud GPU infrastructure and storage to develop robots for industrial work. NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs support simulation and evaluation, while Blackwell platforms support large-scale model training and experimentation. Object storage and a high-performance file system help the team access essential data and archive or purge temporary data. Before adopting Nebius, RoboForce had separate workflows for data collection, simulation, training, evaluation, deployment and inference.

The development workflow combines real-world robotic data with synthetic data generated using NVIDIA Cosmos. NVIDIA OSMO orchestrates distributed training and evaluation workloads. In Nebius’s March 2026 announcement, co-founder Calvin Zhou described generating thousands of scenario variations on Nebius AI Cloud and using one configuration file to run the pipeline end to end. He reported iteration cycles falling from weeks to days. The scenario variations support RoboForce’s robot foundation model for unstructured outdoor environments, including solar farms, construction sites and agricultural fields. Zhou said the workflow allows RoboForce to push hardened robot models to the edge. NVIDIA also names RoboForce as an early user of its Physical AI Data Factory Blueprint on Nebius infrastructure to accelerate model development.

At Nebius Inflection, a RoboForce representative described faster development cycles with the setup. Nebius also reported more than 70% less scenario-generation setup time. Its customer story reports a 70% reduction in setup pipeline time and iteration cycles shortened from months to days. Nebius’s awards page says RoboForce began building on Nebius AI Cloud after winning the 2025 Industrial Robotics award.