NVIDIA and Academic Partners Open-Source CaP-X Under MIT License, Accelerating Robotics Research Access
NVIDIA researcher Jim Fan announced the full open-source release of CaP-X, a project developed in collaboration with UC Berkeley, Stanford, and CMU, under the permissive MIT license.
9. NVIDIA and Academic Partners Open-Source CaP-X Under MIT License, Accelerating Robotics Research Access
NVIDIA researcher Jim Fan announced the full open-source release of CaP-X, a project developed in collaboration with UC Berkeley, Stanford, and CMU, under the permissive MIT license. The release includes the full codebase and accompanying research paper. The project was co-advised by Ken Goldberg, a leading robotics researcher at Berkeley known for foundational work in robot manipulation and learning. The MIT license means anyone can use, modify, and commercialize the work without restriction.
The release matters because CaP-X joins a growing body of NVIDIA-backed robotics and embodied AI research being pushed directly into the open-source commons, a deliberate strategy that seeds the broader ecosystem with tools built on NVIDIA infrastructure. Goldberg's co-advising role signals serious academic credibility, and the tri-university coalition of Berkeley, Stanford, and CMU represents three of the most competitive robotics research pipelines in the world. Startups and labs building on CaP-X will naturally orient their stacks toward NVIDIA hardware, while competing robotics platforms from Boston Dynamics, Figure AI, and Physical Intelligence face a research ecosystem increasingly shaped by NVIDIA's open tooling choices. The losers here are proprietary robotics middleware vendors whose value proposition erodes each time a well-resourced open release lands with an MIT license.
This follows a clear pattern in NVIDIA's research posture: use open-source releases co-branded with elite institutions to establish technical standards before markets fully form. Isaac Lab, GROOT, and now CaP-X form a compounding surface area of influence over how embodied AI gets built. Each MIT-licensed drop is less a gift to the commons than a long-term infrastructure play.
Source: https://twitter.com/DrJimFan/status/2039360925606760690