• 3 min read
Nvidia pulls Japan robot giants into Cosmos
Nvidia says 22 Japanese companies intend to join its Cosmos Coalition, including FANUC and Yaskawa, as it pushes open world models for robotics.

Image: TNW
Nvidia has lined up much of Japan’s industrial robotics sector behind its Cosmos Coalition, with 22 companies named during Jensen Huang’s week in Tokyo. The list includes AIRoA, classmethod, Enactic, FANUC, Fujitsu, GROOVE X, Hitachi, Honda R&D, Kawasaki Heavy Industries, Kubota, Mitsui & Co., Mitsubishi Corp., Mujin, NEC, Preferred Networks, SoftBank Corp., Sony Group, Telexistence, TIER IV, TRON K.K., Turing, and Yaskawa Electric.
In Nvidia’s wording, all of them “intend to join.” No binding commitment has been disclosed, and neither side mentioned any money. Even so, the lineup matters: FANUC and Yaskawa are the world’s two largest industrial robot makers by installed base, and both have spent decades building around proprietary control stacks.
The technical centerpiece is Cosmos 3 Edge, a four-billion-parameter model based on Nvidia’s Nemotron family. It runs on Jetson edge hardware instead of a data center, and Nvidia says developers can adapt it to a specific robot, vehicle, or sensor rig in about a day. The company says it will deploy across RTX GPUs, DGX systems, and the newly announced Jetson T2000 and T3000 modules.
Fujitsu’s collaborative control platform
The clearest project so far comes from Fujitsu, which is leading a collaborative control platform being explored with FANUC, Yaskawa, and Kawasaki. Nvidia says it would bridge digital and physical operations across industries using Cosmos world foundation models, the Isaac robotics platform, Omniverse NuRec libraries, and the Newton physics engine. The goal is to support digital twins, robot learning, and simulation-to-real validation before anything reaches a factory floor.
Other partners are targeting more specific uses:

Recommended reading
Tarit claims microVM speeds 2x past Firecracker
- Kubota is evaluating Cosmos for autonomous agriculture
- Enactic is fine-tuning Nvidia’s Isaac GR00T for elder-care semi-humanoid robots
- Shimizu Corporation is using Metropolis for construction-site safety
- GROOVE X is building Jetson-powered companion robots
- Kawasaki is applying the technology across healthcare, shipbuilding, transportation, aerospace, and energy
“Japan invented modern manufacturing. Now, it has the opportunity to reinvent it for the age of intelligent industries.”
Huang also called physical AI “a once-in-a-generation opportunity” for Japan.
A separate Nvidia release the same day focused on language models. Institute of Science Tokyo built its Swallow open models on Nemotron datasets. SB Intuitions, SoftBank’s generative AI unit, trained its Sarashina series using Nemotron libraries, and Sarashina3 mini has since been adopted by Japan’s Digital Agency. SoftBank Corp. has deployed a Large Telecom Model for autonomous network operations, while NTT DATA used Nvidia’s Japanese personas dataset to augment training for its tsuzumi 2 model. Sakana AI is also integrating Nemotron into its Fugu model-routing platform.
The pitch throughout is sovereignty.
“Every nation and every company should own and control its intelligence infrastructure. Open models make that possible.”
That argument lands awkwardly, since the open models Nvidia is promoting run best on Nvidia hardware. Japan’s interest, though, is more immediate than the rhetoric: Nvidia points to the country’s ageing population and workforce transition, a softer description of the labor shortage manufacturers have been navigating for years. What remains unresolved is whether world models actually shorten the path from robotics demos to real deployment.
Enterprise Editor
Marcus follows the money. He covers enterprise software, cloud architecture, and the tectonic shifts in Big Tech strategy. He translates dense earnings calls and complex M&A activity into actionable insights about where the industry is actually heading. If a tech giant makes a silent pivot, Marcus is usually the first to notice.
via TNW


