Rhoda uses Nebius GPUs and storage to train robotics models from large video collections. In its August 2026 customer interview, the team described doubling its compute with Nebius, using managed Kubernetes and Slurm to manage training, and keeping petabyte-scale data close to the cluster. Rhoda emphasized that compute interconnection matters alongside capacity and chip reliability. Storage infrastructure built by Nebius alongside the cluster allowed the team to focus more on model development and robotics performance.
Rhoda also described testing Nebius infrastructure and considering chip reliability in its decision-making. At Inflection 2026, Rhoda reported that the larger setup let it run more experiments while saving over $1 million that year.
Nebius Inside Rhoda AI: Teaching Robots From Internet-Scale Video
10 Aug 2026 · Nebius · Video
At 2:07, Rhoda describes doubling compute capacity with Nebius. At 2:27 the speaker identifies Kubernetes and Slurm for model training. At 2:47 the team describes petabyte-scale storage beside the cluster.
Nebius Inside Rhoda AI: Teaching Robots From Internet-Scale Video10 Aug 2026 · Nebius · Video · Starts at 2:07At 2:07, Rhoda describes doubling compute capacity with Nebius. At 2:27 the speaker identifies Kubernetes and Slurm for model training. At 2:47 the team describes petabyte-scale storage beside the cluster.
Nebius Inflection 2026 · 20:38
19 Jun 2026 · Nebius · Video
Nebius names Rhoda among its AI-lab customers developing robotics AI and describes applying co-engineering lessons to support these customers.
Nebius Inflection 2026 · 20:3819 Jun 2026 · Nebius · VideoNebius names Rhoda among its AI-lab customers developing robotics AI and describes applying co-engineering lessons to support these customers.
Nebius-commissioned customer-story production
17 Sept 2026 · 175productions.com · Web page
175 Productions describes Rhoda testing Nebius infrastructure and emphasizing compute interconnection alongside capacity and chip reliability. Rhoda uses managed Kubernetes and Slurm for model training. Nebius built storage alongside Rhoda’s compute cluster, allowing the team to focus more on model development and robotics performance.
Nebius-commissioned customer-story production17 Sept 2026 · 175productions.com · Web page175 Productions describes Rhoda testing Nebius infrastructure and emphasizing compute interconnection alongside capacity and chip reliability. Rhoda uses managed Kubernetes and Slurm for model training. Nebius built storage alongside Rhoda’s compute cluster, allowing the team to focus more on model development and robotics performance.
SHLQ226 (1).pdf
18 Oct 2026 · assets.nebius.com · PDF
Nebius names Rhoda among the members of its Customer Advisory Board launched at the June 9 Inflection event.
SHLQ226 (1).pdf18 Oct 2026 · assets.nebius.com · PDFNebius names Rhoda among the members of its Customer Advisory Board launched at the June 9 Inflection event.
Rhoda develops robot foundation models and robotic systems for automotive, manufacturing, logistics and ecommerce customers. Its FutureVision intelligence platform uses a Direct Video Action model trained on internet video and robot data to translate predicted movements into physical actions, adjusting as the robot observes its surroundings. Applications include processing returned goods, handling components on automotive assembly lines and collapsing reusable shipping containers. Rhoda also builds a robot platform with custom actuators, integrated brakes and safety-rated vision for industrial material handling.
Rhoda introduces tactile sensor and robot model that manipulates objects by touch
Rhoda unveiled a tactile sensor integrated with its multimodal Direct Video-Action model. Demonstrations showed a robot identifying and retrieving textured objects with its cameras blocked.
Rhoda uses Nebius GPUs and storage to train robotics models from large video collections. In its August 2026 customer interview, the team described doubling its compute with Nebius, using managed Kubernetes and Slurm to manage training, and keeping petabyte-scale data close to the cluster. Rhoda emphasized that compute interconnection matters alongside capacity and chip reliability. Storage infrastructure built by Nebius alongside the cluster allowed the team to focus more on model development and robotics performance.
Rhoda also described testing Nebius infrastructure and considering chip reliability in its decision-making. At Inflection 2026, Rhoda reported that the larger setup let it run more experiments while saving over $1 million that year.
Nebius Inside Rhoda AI: Teaching Robots From Internet-Scale Video
10 Aug 2026 · Nebius · Video
At 2:07, Rhoda describes doubling compute capacity with Nebius. At 2:27 the speaker identifies Kubernetes and Slurm for model training. At 2:47 the team describes petabyte-scale storage beside the cluster.
Nebius Inside Rhoda AI: Teaching Robots From Internet-Scale Video10 Aug 2026 · Nebius · Video · Starts at 2:07At 2:07, Rhoda describes doubling compute capacity with Nebius. At 2:27 the speaker identifies Kubernetes and Slurm for model training. At 2:47 the team describes petabyte-scale storage beside the cluster.
Nebius Inflection 2026 · 20:38
19 Jun 2026 · Nebius · Video
Nebius names Rhoda among its AI-lab customers developing robotics AI and describes applying co-engineering lessons to support these customers.
Nebius Inflection 2026 · 20:3819 Jun 2026 · Nebius · VideoNebius names Rhoda among its AI-lab customers developing robotics AI and describes applying co-engineering lessons to support these customers.
Nebius-commissioned customer-story production
17 Sept 2026 · 175productions.com · Web page
175 Productions describes Rhoda testing Nebius infrastructure and emphasizing compute interconnection alongside capacity and chip reliability. Rhoda uses managed Kubernetes and Slurm for model training. Nebius built storage alongside Rhoda’s compute cluster, allowing the team to focus more on model development and robotics performance.
Nebius-commissioned customer-story production17 Sept 2026 · 175productions.com · Web page175 Productions describes Rhoda testing Nebius infrastructure and emphasizing compute interconnection alongside capacity and chip reliability. Rhoda uses managed Kubernetes and Slurm for model training. Nebius built storage alongside Rhoda’s compute cluster, allowing the team to focus more on model development and robotics performance.
SHLQ226 (1).pdf
18 Oct 2026 · assets.nebius.com · PDF
Nebius names Rhoda among the members of its Customer Advisory Board launched at the June 9 Inflection event.
SHLQ226 (1).pdf18 Oct 2026 · assets.nebius.com · PDFNebius names Rhoda among the members of its Customer Advisory Board launched at the June 9 Inflection event.
Rhoda develops robot foundation models and robotic systems for automotive, manufacturing, logistics and ecommerce customers. Its FutureVision intelligence platform uses a Direct Video Action model trained on internet video and robot data to translate predicted movements into physical actions, adjusting as the robot observes its surroundings. Applications include processing returned goods, handling components on automotive assembly lines and collapsing reusable shipping containers. Rhoda also builds a robot platform with custom actuators, integrated brakes and safety-rated vision for industrial material handling.
Rhoda introduces tactile sensor and robot model that manipulates objects by touch
Rhoda unveiled a tactile sensor integrated with its multimodal Direct Video-Action model. Demonstrations showed a robot identifying and retrieving textured objects with its cameras blocked.
