NVIDIA announced that Japanese enterprises, startups and research institutions are building industry-specialized AI models and applications with NVIDIA Nemotron open models, data and libraries, which the company said is "accelerating the development of AI tailored to Japan's language, industries and workforce." NVIDIA described the participants in its release as "leading Japanese enterprises, startups and research institutions."
NVIDIA founder and CEO Jensen Huang framed the strategy around technological sovereignty, stating: "Every nation and every company should own and control its intelligence infrastructure. Open models make that possible."
According to NVIDIA, the Nemotron models underpinning these projects are distributed openly: they are available on Hugging Face, ModelScope, OpenRouter and build.nvidia.com as NVIDIA NIM microservices, and through NVIDIA Cloud Partners, inference platforms and cloud service providers. This open distribution model is what allows the range of organizations described below — from a national research university to startups and industrial conglomerates — to build on the same underlying models.
Per NVIDIA's release, Institute of Science Tokyo developed its Swallow family of open foundation models using NVIDIA Nemotron datasets and the NVIDIA NeMo software stack, applying them specifically for continual pretraining and post-training. This academic effort is one of the examples NVIDIA cited to support its broader claim that Nemotron is being used across Japan's research institutions, alongside enterprise and startup deployments, to develop AI tailored to Japan's language and industries.
Two startups detailed in NVIDIA's announcement illustrate different applications of Nemotron. SB Intuitions Corp., SoftBank Corp.'s generative AI research subsidiary, trained its Sarashina series of homegrown generative AI models using NVIDIA Nemotron, including the NVIDIA NeMo RL and Megatron-LM libraries. Notably, NVIDIA said Sarashina3 mini "has been selected by Japan's Digital Agency for use in specialized AI use cases" — a government adoption point distinct from the other examples in the release.
Stockmark, according to NVIDIA, released a specialized Japanese-language document-understanding model based on the NVIDIA Nemotron 3 Nano Omni model. The company is also developing enterprise knowledge applications using NVIDIA NeMo Retriever and the Nemotron-Personas-Japan dataset, and NVIDIA said Stockmark serves customers across Japan's manufacturing, energy and chemical industries through Japan's Generative AI Accelerator Challenge national project.
NVIDIA's release lists four large Japanese enterprises using Nemotron for distinct purposes. AI and robotics startup avatarin — grouped by NVIDIA alongside the enterprise deployments — is using NVIDIA Nemotron open models and NVIDIA NeMo to develop Japanese-language speech and reasoning capabilities for enterprise AI agents. NVIDIA HGX B300 systems provide the private AI infrastructure NVIDIA said enables those agents to "securely analyze customer conversations and access enterprise knowledge," while NVIDIA Jetson powers edge AI capabilities including digital avatar systems being deployed at airports and other locations across Japan.
ENEOS Holdings is using NVIDIA Nemotron open models with the NVIDIA AI-Q Blueprint and NVIDIA ALCHEMI NIM microservices to advance agentic AI workflows for energy and materials R&D, per NVIDIA.
NTT DATA, an operating subsidiary of NTT, used NVIDIA Nemotron-Personas-Japan to augment training data for its proprietary tsuzumi 2 model, which NVIDIA said improved question-answering accuracy and responses to questions requiring additional knowledge. NTT DATA is also looking to deploy a multi-agent framework harnessing NVIDIA Agent Toolkit, including Nemotron, to route tasks to the models NVIDIA said would be best suited for each job.
Hitachi is developing physical AI technologies using NVIDIA Nemotron and NVIDIA Cosmos open models combined with its own information technology (IT) and operational technology (OT) domain knowledge. As part of a multi-agent orchestration platform, NVIDIA said these technologies are designed to connect and coordinate IT and OT operations, aiming to help transform enterprise-scale business processes across complex workflows.
Sakana AI is collaborating with NVIDIA to integrate NVIDIA Nemotron into its Fugu model-orchestration platform, according to the announcement. NVIDIA said this expands the range of AI models Fugu can orchestrate to dynamically select the best model for each task in agentic AI workflows — an approach made possible by Nemotron's availability as an open model through Hugging Face, ModelScope, OpenRouter and build.nvidia.com's NIM microservices rather than a single closed deployment path.
| Organization | NVIDIA Technology Cited | Stated Application |
|---|---|---|
| Institute of Science Tokyo | Nemotron datasets, NeMo software stack | Swallow foundation models (continual pretraining/post-training) |
| SB Intuitions Corp. | Nemotron, NeMo RL, Megatron-LM | Sarashina series; Sarashina3 mini selected by Japan's Digital Agency |
| Stockmark | Nemotron 3 Nano Omni, NeMo Retriever, Nemotron-Personas-Japan | Japanese document understanding; manufacturing/energy/chemical clients via Generative AI Accelerator Challenge |
| avatarin | Nemotron, NeMo, HGX B300, Jetson | Japanese-language speech/reasoning agents; airport digital avatars |
| ENEOS Holdings | Nemotron, AI-Q Blueprint, ALCHEMI NIM microservices | Agentic AI workflows for energy and materials R&D |
| NTT DATA | Nemotron-Personas-Japan, Agent Toolkit | tsuzumi 2 training data augmentation; multi-agent framework |
| Hitachi | Nemotron, Cosmos | IT/OT physical AI orchestration platform |
| Sakana AI | Nemotron | Fugu model-orchestration platform |
Taken together, the organizations NVIDIA named span three distinct layers of Japan's AI ecosystem: a research institution (Institute of Science Tokyo's Swallow models), startups building proprietary systems (SB Intuitions, Stockmark, avatarin, Sakana AI), and large enterprises embedding Nemotron into operational workflows (ENEOS Holdings, NTT DATA, Hitachi). Two data points in NVIDIA's release point toward institutional adoption beyond private industry: SB Intuitions' Sarashina3 mini was selected by Japan's Digital Agency, and Stockmark's document-understanding work is tied to Japan's Generative AI Accelerator Challenge national project — meaning Nemotron-based models are already connected to government-affiliated channels, not only corporate deployments. At the same time, Sakana AI's integration of Nemotron into its Fugu orchestration platform, alongside NVIDIA's confirmation that Nemotron is distributed openly through Hugging Face, ModelScope, OpenRouter and build.nvidia.com, indicates that per NVIDIA's own framing, Nemotron is positioned as one interchangeable option among multiple models an orchestration layer can select from, rather than a single mandated stack.
According to NVIDIA, Japan's Digital Agency selected SB Intuitions' Sarashina3 mini — a model trained using NVIDIA Nemotron, NeMo RL and Megatron-LM — for use in specialized AI use cases.
NVIDIA states that Nemotron models are available on Hugging Face, ModelScope, OpenRouter and build.nvidia.com as NVIDIA NIM microservices, and through NVIDIA Cloud Partners, inference platforms and cloud service providers.
Per NVIDIA's announcement, Institute of Science Tokyo used NVIDIA Nemotron datasets and the NVIDIA NeMo software stack for continual pretraining and post-training of its Swallow family of open foundation models.
According to Intel's newsroom announcement, the company is investing €5 billion ($5.7 billion) to expand its Leixlip campus in Ireland, producing Intel Xeon 6 and next-generation Xeon processors on the Intel 3 node. Intel says the capital programme began earlier this year and builds on more than €30 billion invested in Ireland since 1989.
According to TechCrunch and VentureBeat, Meta released Muse Code — a beta terminal-based AI coding agent for large repositories — alongside the Muse Spark 1.2 model on August 5, 2026. Muse Code installs with a single command, runs persistent background agents, and is entirely proprietary, putting Meta in direct competition with Anthropic's Claude Code and OpenAI's Codex.
NVIDIA has moved Alpamayo 2 Super into commercial use, according to NVIDIA's official blog. The autonomous-driving reasoning model runs on Cosmos 3 Super Reasoner, holds three times the parameters of its 10-billion-parameter predecessors, and ships under the OpenMDW-1.1 license, which ITHome reports NVIDIA has extended across the entire Alpamayo model family for fine-tuning and commercial redistribution.