NVIDIA has spent years building the chips that power many of the world’s best known AI models. Now, it wants developers to use more of its own AI models too.
The company announced that Japanese businesses, startups and research organisations are building applications using its open source Nemotron models, data and software libraries. The announcement centres on Japan, where organisations want AI that understands the country’s language, workplaces and public services.
The news is quite interesting to me, especially since NVIDIA’s hardware powers a long list of AI models available through its platform, from OpenAI’s gpt oss and Meta’s Llama to Google’s Gemma, Microsoft’s Phi, Alibaba’s Qwen, Moonshot AI’s Kimi and DeepSeek. Now NVIDIA is giving developers another option from its own catalogue.
Why Is NVIDIA Putting More Focus Onto Nemotron?
NVIDIA says open models give organisations more control over how AI works in their businesses. Instead of relying entirely on closed systems, companies can inspect the models, adapt them for their own work and run them where their data is stored.
Jensen Huang, founder and CEO of NVIDIA, said, “Every nation and every company should own and control its intelligence infrastructure. Open models make that possible. They give countries, enterprises and researchers the freedom to inspect, improve, adapt, secure and deploy AI for their own needs. Together with Japan’s AI leaders, we are advancing an open AI ecosystem that accelerates discovery, strengthens national capability and ensures every society can participate in and benefit from the AI revolution.”
Japan gives NVIDIA an opportunity to demonstrate that vision. NVIDIA said an ageing population and workforce transition have increased demand for AI that understands Japanese language and business needs.
The Institution of Science Tokyo has built its Swallow family of foundation models using NVIDIA Nemotron datasets and the NVIDIA NeMo software stack. NVIDIA said businesses are already adapting Swallow for financial document translation and asset management report generation.
SoftBank subsidiary SB Intuitions trained its Sarashina family of generative AI models using NVIDIA Nemotron, NVIDIA NeMo RL and Megatron LM. NVIDIA said Sarashina3 mini has already been selected for specialised AI work at Japan’s Digital Agency.
More from Artificial Intelligence
- What’s The Difference Between Open Weight And Open Source AI?
- NVIDIA’s AI Tool Can Spot Deepfake Videos – Is The Company Starting Too Many Projects At Once?
- Frontier AI Or Budget Models – Which Strategy Makes Sense For Your Business?
- AI Is Supposedly “Replacing” Writers. So Why Is Anthropic Hiring Editors For $295,000 A Year?
- Should AI Literacy Be Taught Alongside Math And English?
- Should AI Have An Appreciation Day When Other Industries Don’t?
- The Rise Of AI-Powered Product Governance: From Institutional Knowledge To Scalable Decision-Making
- ChatGPT Can Now Handle Your Company’s Web Domain Admin – Here’s How
What Does This Mean For Competition?
NVIDIA already has a unique place in AI because many well known models depend on its hardware and software to run efficiently.
On NVIDIA’s own AI model platform, developers can access DeepSeek, Google’s Gemma, OpenAI’s gpt oss, Moonshot AI’s Kimi, Meta’s Llama, Microsoft’s Phi and Alibaba’s Qwen. NVIDIA has also worked to optimise many of these models for its Blackwell, Hopper and RTX hardware.
Nemotron therefore enters a marketplace where NVIDIA already supports competing models. Instead of asking developers to leave those models behind, the company is placing its own family beside them.
NVIDIA says Nemotron is available in Nano, Super and Ultra versions. Nano is designed for cost efficiency, Super is designed for a mix of accuracy and computing resources, and Ultra is designed for maximum accuracy. The models are released under an open licence, giving organisations commercial freedom and control over their data.
That also means developers already using NVIDIA infrastructure can experiment with Nemotron without leaving the platform they already know.
How Are Companies Already Using It?
Japanese businesses are already building applications in healthcare, robotics, telecommunications, manufacturing and energy.
Avatarin is developing Japanese language speech and reasoning systems for enterprise AI agents. ENEOS Holdings is using Nemotron together with NVIDIA software for energy and materials research, helping researchers search technical documents, understand images and language, and screen molecules through simulations.
NTT DATA used the Nemotron Personas Japan dataset to train its proprietary tsuzumi 2 model, improving question answering and responses that need extra knowledge. The company also plans a multi agent framework that routes work to the most suitable AI model for each task.
Hitachi is building physical AI technologies using Nemotron and NVIDIA Cosmos models for business and operational systems. Sakana AI is integrating Nemotron into its Fugu model routing platform, giving developers another model that can be selected automatically according to the work being carried out.
That example may say the most about NVIDIA’s strategy; rather than asking developers to choose one AI model forever, the company is building tools where Nemotron becomes one more choice in a growing collection of models. Since many of those models already run on NVIDIA hardware, the company now wants its own software to become another regular choice for developers building AI applications.
