Nvidia AI Push Names Banks, Carmakers And RIKEN
Tech Wire Asia reported that Nvidia used a country visit by Jensen Huang to connect Nemotron open models, DGX B200 systems and Blackwell supercomputers with local banks, carmakers, hospitals and research groups. The report said pilot-to-production conversion, audited benchmark results and chip-order volumes remain outside the public record.
Nvidia is positioning Japan as its clearest example of sovereign AI: locally adapted models, on-premises systems and applications built around Japanese data, all running on Nvidia hardware.
The campaign took shape during Jensen Huang’s first visit to the country in about nine months and brought together banks, carmakers, medical-imaging groups, research institutes and robotics companies.
Local Models And Deployment Stack
The software layer is Nvidia’s Nemotron family of open-weight models, released with datasets and training material that companies can fine-tune on domestic data and deploy inside their own borders.
The approach makes the models locally governable while tying their operation to Nvidia’s GPU systems.
The Institute of Science Tokyo used Nemotron data to train its Swallow foundation models.
SB Intuitions, SoftBank’s subsidiary, used Nemotron libraries for its Sarashina series, while Sakana AI is integrating Nemotron into its Fugu model-routing platform.
NTT DATA, Hitachi and ENEOS are building Japanese-language applications on top of the models.
Huang has described sovereign AI as a strategy spanning more than 20 countries.
In Japan, the pitch is also tied to an ageing population and shrinking workforce, with AI and robotics presented as tools for sustaining productivity as labour becomes scarcer.
Banks, Research And Healthcare Systems
Mizuho plans what it expects to be the largest on-premises AI factory in Japan’s banking sector, beginning with Nvidia DGX B200 systems.
Keeping the factory inside the bank is intended to let AI agents work with sensitive data without that information leaving the building.
SMBC Group’s Japan Research Institute has deployed a Nemotron-based AI factory, while Rakuten Bank is building transaction foundation models on tens of millions of banking and card accounts held across the group.
RIKEN is bringing two supercomputers built on Nvidia’s Blackwell chips online, including a quantum-HPC system linked to on-premises quantum processors.
Nvidia has also characterized Japan as the first international partner in the US government’s Genesis Mission, an AI-for-science initiative.
In healthcare, the Tokyo-1 drug-discovery consortium run by Xeureka includes Astellas, Daiichi Sankyo, Ono Pharmaceutical and, since April, Eisai.
The group is running screening and generative-design workloads on Nvidia’s BioNeMo platform.
SyntheticGestalt unveiled two molecular models that it says lead public drug-discovery benchmarks, although that claim comes from the company’s own testing rather than independent evaluation.
Robotics And Industrial Operations
Kawasaki Heavy Industries is developing surgical-support, nursing and transport robots.
Toyota is extending an existing partnership to build driver-assistance systems on Nvidia’s DRIVE platform, run factory simulations in Omniverse and train a vision-language model through its Woven subsidiary.
Nvidia’s Metropolis toolkit is being used by OMRON, Fujitsu, Hitachi and Shimizu for factory inspection, rail maintenance and construction-site safety.
Canon and Fujifilm have begun shipping next-generation CT systems built on Nvidia GPUs.
Efficiency figures attached to these projects include a 15% cut in rail maintenance and energy costs, but those numbers come from vendors and their partners.
China Controls And The Deployment Boundary
The Japan campaign is unfolding as tighter US export controls limit Nvidia’s access to China’s data-centre market.
Washington tightened the controls again in May to cover the company’s Blackwell line.
On July 14, Nvidia introduced a stricter compliance regime for Asian customers.
The reported whitelist disqualified more than half of past buyers and added end-user scrutiny in Singapore, Malaysia and Japan to reduce the risk that chips would be rerouted into China.
China’s LineShine system meanwhile reclaimed the top spot on the June TOP500 list.
The Japanese partner list shows how Nvidia is building an allied ecosystem around local models and domestic data.
Its production scale remains uneven: Mizuho’s on-premises commitment and the CT systems already shipping are among the strongest deployment signals, while some robotics projects and benchmark claims remain closer to demonstration or company-reported results.
Japan’s sovereign AI strategy, for now, still runs on Nvidia’s chips.
