Nvidia Launches Cosmos 3 Edge World Model for Physical AI, Forms Japan Industrial Coalition
Nvidia released Cosmos 3 Edge, a world model designed for robots and vision AI agents to perceive and navigate physical environments in real time. The company announced partnerships with Japanese industrial giants including Fujitsu, Hitachi, and Kawasaki Heavy Industries as part of its physical AI expansion.
Nvidia Launches Cosmos 3 Edge World Model for Physical AI, Forms Japan Industrial Coalition
Nvidia released Cosmos 3 Edge on Wednesday, a world model designed to help robots and vision AI agents perceive and navigate physical environments in real time. The model follows the company's Cosmos 3 launch in May.
World models are systems that can learn from a wider range of inputs compared to large language models, according to Nvidia. Cosmos 3 Edge is specifically designed for edge deployment in physical AI applications.
Japan Industrial Coalition
During CEO Jensen Huang's two-day Japan visit, Nvidia announced formation of a physical AI coalition that Fujitsu, Hitachi, and Kawasaki Heavy Industries intend to join. The expansion aims to capitalize on Japan's manufacturing expertise.
"The next frontier of AI is in the physical world, and this is a once-in-a-generation opportunity for Japan," Huang said in a statement. "Japan invented modern manufacturing. Now, it has the opportunity to reinvent it for the age of intelligent industries."
Japan's AI market is expected to reach $27.9 billion by 2029, according to the International Trade Administration. The expansion follows Microsoft's $10 billion investment in Japan announced earlier this year for AI infrastructure and cybersecurity.
Healthcare and Drug Discovery Partnerships
Nvidia extended its reach into Japan's healthcare sector through drug discovery and medical robotics initiatives. The company highlighted Tokyo-1, an AI drug discovery consortium operated by Xeureka, a Mitsui subsidiary, which runs on the Nvidia BioNeMo Agent Toolkit.
Major Japanese pharmaceutical companies including Astellas Pharma, Daiichi Sankyo, and Ono Pharmaceutical are using Nvidia's biology toolkit to streamline workflows, according to the company. The Tokyo-1 platform has expanded since its initial 2023 announcement.
Nvidia also announced a partnership with Kawasaki Heavy Industries for industrial automation, though specific details were not disclosed.
What This Means
Nvidia's Japan push represents a strategic bet on physical AI—systems that interact with the real world—beyond pure software applications. By partnering with Japan's manufacturing and pharmaceutical giants, Nvidia is positioning its hardware and software stack as infrastructure for robotics, automated factories, and drug discovery at scale. The timing aligns with Japan's government push for AI adoption and comes as U.S. tech companies race to establish footholds in Asia's second-largest economy. The success of these partnerships will test whether world models can deliver practical value in industrial settings where reliability and precision are non-negotiable.
Related Articles
Nvidia Claims Groq 3 LPX Hits 3,400 Tokens/Sec, 4x Cerebras — But Needs 64 Chips to Do It
Nvidia's new Groq 3 LPX inference accelerator hit 3,400 tokens per second on Gemma 4 31B, a figure the company says is four times faster than Cerebras. Experts note the benchmark uses at least 64 LPX chips versus Cerebras' one or two accelerators, making the comparison far less clean than it appears.
Tencent Open-Sources Hy4 Preview: 770B-Parameter MoE Model with 1M-Token Context
Tencent's Hy Team has open-sourced Hy4 preview, a 770-billion-parameter Mixture-of-Experts model with 49 billion activated parameters and a 1-million-token context window. The model is available under Apache 2.0 alongside an FP8-quantized variant, with Tencent claiming it beats GLM 5.3 and Kimi K3 on internal engineering evaluations.
Tencent Releases Hy4 Preview: 770B-Parameter MoE Model with 1M Context for Coding Agents
Tencent has released Hy4 preview, a mixture-of-experts model with 770B total parameters and 49B active parameters, targeting coding agents and multi-step tool-use workflows. The model ships with a 1 million token context window and is priced at $0.834 per 1M input tokens and $2.501 per 1M output tokens.
Google Launches Gemini 3.5 Transcribe with 4.0% Word Error Rate Across 85 Languages
Google has released Gemini 3.5 Transcribe, a speech-to-text model that automatically detects 85 languages, removes filler words, and corrects misspoken phrases. The company claims a 4.0 percent word error rate for streaming audio and 70 percent lower latency than its predecessor, Chirp 3.
Comments
Loading...