N※N

Japan Unveils NVIDIA-Driven Supercomputers for Next-Gen HPC

80

  //science-technology.news-articles.net/content/2 .. idia-driven-supercomputers-for-next-gen-hpc.html
  Published in Science and Technology on Tuesday, November 18th 2025 at 11:28 GMT by Interesting Engineering

• 🞛 This publication is a summary or evaluation of another publication
• 🞛 This publication contains editorial commentary or bias from the source

2025-11-18 224 x 224 / 9280 Bytes

2025-11-18 179 x 224 / 12524 Bytes

2025-11-18 179 x 224 / 5480 Bytes

2025-11-18 179 x 224 / 5163 Bytes

NVIDIA GPUs Power Japan’s New Generation of Supercomputers

In a move that signals a major shift in Japan’s high‑performance computing (HPC) landscape, the country has unveiled plans for a new wave of supercomputers that will be powered by NVIDIA’s cutting‑edge graphics processing units (GPUs). The announcement, made by Japan’s Ministry of Education, Culture, Sports, Science and Technology (MEXT) and the Institute for Advanced Simulation (IAS), highlights a partnership with NVIDIA that will give Japanese researchers unprecedented computational muscle for a range of scientific and industrial challenges.

From Fugaku to the Future

The context of this announcement is crucial. Japan’s previous flagship supercomputer, Fugaku, built by Fujitsu and the RIKEN Advanced Institute for Computational Science, held the world’s top spot on the TOP500 list from 2020 to 2023. Fugaku was a CPU‑centric machine that used ARM‑based A64FX processors. While it performed well for many workloads, NVIDIA has been pushing a GPU‑centric model that is becoming the new benchmark for AI‑driven and data‑intensive workloads.

MEXT’s new supercomputing roadmap explicitly states that the next generation of machines will pivot to a GPU‑accelerated architecture. This will allow Japan to keep pace with global leaders such as the United States, China, and Europe, where NVIDIA GPUs dominate the top tier of the TOP500. The announcement referenced a series of papers published in Nature and Science that demonstrate the superior speed‑up NVIDIA’s H100 Tensor Core GPUs bring to complex simulations—especially for quantum chemistry, materials science, and climate modeling.

The Two New Behemoths: “Shinshu” and “Kōkō”

The article identified two new supercomputers that will make up Japan’s next generation of HPC power:

1. Shinshu – to be hosted at the Institute for Advanced Simulation (IAS) in Nagano. It is expected to use a hybrid architecture of NVIDIA H100 GPUs and Fujitsu’s A64FX CPUs, with a projected peak performance of more than 2 petaflops. The system will support “AI‑first” workloads, from protein folding to autonomous robotics.

2. Kōkō – a dedicated machine for the RIKEN Center for Computational Science (CCS). This system will also rely on NVIDIA’s H100 GPUs and will focus on climate and weather modeling, enabling more accurate, high‑resolution simulations of extreme weather events.

Both projects are set to be commissioned in the second half of 2025. The article linked to a detailed timeline published by the Japanese HPC consortium, which shows that each supercomputer will undergo a phased rollout: initial commissioning, performance validation against benchmarks (LINPACK, Graph500), and final certification for use in national research projects.

NVIDIA’s Role: GPUs, Software, and Support

NVIDIA’s contribution goes beyond just supplying GPUs. The company has also offered software stack support, including the CUDA Toolkit, NVIDIA RAPIDS, and NVIDIA DGX‑A100 pre‑configured systems. These tools are designed to simplify the migration of legacy codes to a GPU‑centric environment. The article cited a partnership agreement that includes NVIDIA’s GPU Enterprise Software License, which ensures that Japanese researchers can use the full suite of GPU‑accelerated libraries without additional licensing overhead.

Furthermore, NVIDIA’s NVIDIA AI Enterprise platform will be integrated into the training and operation of the supercomputers. This includes GPU‑optimized versions of TensorFlow, PyTorch, and the DeepSpeed library, allowing researchers to accelerate deep‑learning workloads that would have otherwise taken weeks on CPU clusters.

Funding and Economic Impact

Japan’s investment of roughly ¥12 billion (≈US$100 million) per supercomputer is part of a broader “HPC 2030” initiative. The initiative aims to make Japan a global leader in high‑performance computing for 2030. The article linked to the official MEXT budget brief, which details how the funds will cover hardware procurement, facility upgrades, and the training of hundreds of HPC professionals over the next decade.

Economic analysts quoted in the piece estimate that each supercomputer will generate up to ¥3 billion (≈US$25 million) in annual economic activity by driving advances in pharmaceuticals, energy storage, and digital twin technologies. These figures underscore Japan’s strategy of leveraging HPC to create high‑value jobs and stimulate R&D that can be commercialized both domestically and abroad.

Scientific Applications and International Collaboration

The article highlighted several scientific projects that will benefit from the new GPUs:

• Climate science: Higher‑resolution Earth system models will be run at sub‑kilometer scales, improving predictions of typhoon paths and sea‑level rise.
• Materials science: Quantum‑mechanics calculations using density functional theory will be accelerated by a factor of ten, speeding up the discovery of new battery materials.
• Biological research: Protein‑folding simulations will be carried out on a scale that was previously only achievable with supercomputers in the United States or China.

The piece also mentioned international collaborations. For example, Japan is coordinating with the European Centre for Medium‑Range Weather Forecasts (ECMWF) to exchange best practices in GPU‑accelerated climate modeling. Additionally, the article cited a joint research agreement with NVIDIA’s “Global Founders Program,” which will allow select Japanese research groups to co‑develop GPU‑optimized simulation codes.

Challenges and Outlook

While the article paints an optimistic picture, it does not shy away from the challenges that lie ahead. Integrating a new GPU architecture requires significant software refactoring, which can be time‑consuming. Additionally, the power consumption of H100 GPUs—though more efficient than previous generations—will demand upgrades to cooling infrastructure. The article referenced a technical note from the Tokyo Electric Power Company (TEPCO) outlining the need for increased renewable energy supply to sustain the new supercomputers’ energy budget.

Nevertheless, the consensus among experts quoted in the article is that the move to NVIDIA GPUs positions Japan to compete more effectively on the global HPC stage. In the words of Dr. Haruki Yoshida, director of the IAS, “By adopting NVIDIA’s cutting‑edge GPU technology, we are not just catching up—we’re setting the pace for the next wave of scientific discovery.”

Where to Learn More

The article concludes by pointing readers toward several resources for deeper dives:

• NVIDIA’s official HPC page (https://developer.nvidia.com/hpc) for detailed specifications of the H100 Tensor Core GPUs and the associated software ecosystem.
• The TOP500 website’s entry for the latest ranking of the new Japanese supercomputers, once they become operational.
• The RIKEN Center for Computational Science blog, which will provide weekly updates on the build and commissioning phases of the Kōkō system.

In summary, Japan’s decision to power its next generation of supercomputers with NVIDIA GPUs marks a watershed moment in the country’s research infrastructure. By marrying Japan’s robust CPU design heritage with NVIDIA’s GPU acceleration, the nation is poised to tackle some of the most pressing scientific questions of our time, from climate change to next‑generation energy solutions. The coming years will reveal how quickly and effectively these powerful machines translate raw computational horsepower into tangible breakthroughs—an outcome that could redefine the global balance of supercomputing prowess.