Rhoda AI is a physical-AI company building foundation models for robotic intelligence, with the explicit goal of moving robots beyond curated laboratory demonstrations into messy real-world environments like factories, warehouses, and e-commerce fulfillment centers. After roughly 18 months in stealth, it launched publicly in March 2026 by unveiling FutureVision, its intelligence layer for robots, alongside a large funding round that immediately made it one of the most-watched names in the robotics-foundation-model wave of 2026.
At the heart of Rhoda's approach is a proprietary Direct Video Action (DVA) model. Unlike conventional vision-language-action pipelines or open-loop planners that generate a plan and execute it blindly, DVA bridges perception and control directly and updates behavior continuously as conditions change, enabling physics-aware control in real time. The training recipe is distinctive: extensive pre-training on more than one million web-scale videos gives the model broad physical priors, while only one to ten hours of task-specific trajectory data is needed to adapt it to a new job — a data efficiency that addresses the generalization problems that plague many robot-learning systems.
FutureVision powers Rhoda's own systems today and is intended over time to be licensed to partners across different robot hardware and software platforms, positioning Rhoda as an intelligence provider rather than purely a hardware company. The company has demonstrated practical industrial tasks including returns processing, bearing decanting, and box breakdown, targeting automotive, manufacturing, logistics, and e-commerce customers where labor-intensive, variable tasks are hard to automate with traditional robotics.
Rhoda is led by CEO and co-founder Jagdeep Singh, a serial deep-tech founder, with Chief Science Officer Eric Ryan Chan — a Stanford researcher in computer vision and generative modeling who previously worked as a generative-model architect at World Labs — and Stanford professor Gordon Wetzstein, head of the Computational Imaging Lab, among its scientific leadership. This pedigree in generative modeling and computational imaging underpins its video-first thesis.
The company exited stealth with a $450 million Series A at a reported $1.7 billion valuation, backed by Capricorn Investment Group, Khosla Ventures, Leitmotif, Matter Venture Partners, Mayfield, Premji Invest, Prelude Ventures, Temasek, and Xora, with Silicon Valley figures such as John Doerr also participating. The capital funds development and industrial deployment of FutureVision. Rhoda's bet — that learning from internet-scale video is the fastest route to generalizable robot control — places it at the center of one of AI's most consequential frontiers, with the open question being how reliably video-trained policies transfer to the physical edge cases of real industrial work.