Rhoda AI
Palo Alto robotics company building a video-prediction robot control model (DVA, FutureVision) and a wheeled 25 kg-payload robot; USD 450M Series A in 2026.
Rhoda AI is a robotics company based in Palo Alto, California, that develops a video-based robot control model and a wheeled robot platform to run it. It came out of stealth on 10 March 2026 after about 18 months of work incubated at Khosla Ventures, announcing a USD 450 million Series A led by Premji Invest. Other named investors include Khosla Ventures, Temasek, Mayfield, Capricorn Investment Group, Prelude Ventures, Leitmotif, Matter Venture Partners, Xora and John Doerr. Dealroom reported a valuation of about USD 1.7 billion. Jagdeep Singh, who previously founded the battery company QuantumScape, is co-founder and CEO.
The company's core technology is the Direct Video Action (DVA) model. Instead of mapping camera frames straight to actions, DVA predicts future states as video, converts them into robot actions, and re-observes the scene every few hundred milliseconds in a closed loop. The model is pre-trained on large volumes of web video and then post-trained on robot trajectory data; Rhoda says new tasks can need as little as about 10 hours of teleoperation data. It keeps a long history of frames as memory, which lets it handle tasks with visually ambiguous states and follow a single human demonstration without retraining. The intelligence layer built on DVA is called FutureVision, which Rhoda says it runs on its own systems today and expects to license to other robot hardware and software makers over time. In October 2026 the company added a tactile sensor whose measurements feed into the multimodal DVA model.
Rhoda also builds its own wheeled robot platform. Its website lists custom actuators, a 25 kg rated and 40 kg peak payload, brakes in every actuator, safety-rated vision and a design life of three years of continuous operation at rated payload. Published demonstrations cover returns processing in logistics, decanting 10 kg boxes of bearings with tab removal and transparent bags for an automotive use case, and collapsing 50-pound heavy-duty boxes with variable debris for Contico. The company lists automotive, manufacturing, logistics and e-commerce as target verticals. No customers are named publicly, and press coverage mentions only unnamed pilots and a high-volume manufacturing evaluation.
Primary type & automation activities this supplier delivers:
FutureVision
- Launch Year: 2026
Video-predictive robotic intelligence platform for autonomous handling of industrial manipulation tasks.
Direct Video-Action Model (DVA)
- Model: DVA
- Launch Year: 2026
Causal video model that frames robot policy learning as video generation, released as research in March 2026.
Rhoda robot platform
Wheeled manipulation robot with custom actuators built to run Rhoda's video-trained policies in factories.
More related news
- 1
- 2
- 3
- 4
- 5
- 6
Contact Rhoda AI
WEBSITE
https://www.rhoda.ai/HEADQUARTERS
United States
Company Facts
Founded
2024
Primary Role
-
Company Size
-
Primary Region
-
Annual Sales
-
Funding Stage
-
Funding Total
-
Keywords
