Japan’s Fugakunext Supercomputer Set to Redefine Global AI and Scientific Discovery

Japan’s Fugakunext: Setting a New Standard for Supercomputing and AI Research

Japan is pushing the limits of supercomputing with the launch of its latest project: fugakunext. Supported by a $750 million investment, this next-generation machine is expected to outperform existing systems by a factor of 1,000. Developed by RIKEN and Fujitsu, this supercomputer will elevate Japan’s capacity in scientific research and artificial intelligence on a global scale.

A Successor to Fugaku

Fugakunext builds on the success of Fugaku, which debuted in 2020 as one of the world’s fastest supercomputers. Fugaku reached a peak performance of 442 petaFLOPS and played a vital role in COVID-19 pandemic modeling. Its achievements secured Japan a top spot in the Top500 supercomputer rankings, encouraging early planning for its successor.

Since 2022, Japan has invested in feasibility studies to explore the possibility of a zetta-scale supercomputer—one capable of delivering performance 1,000 times greater than current leaders. These studies, involving four research teams, will conclude by March 2024, guiding the design and development of fugakunext.

Technical Ambitions and Innovations

The goal of fugakunext is to reach zetta-scale computing speeds, surpassing the performance of US systems like the Frontier supercomputer. At its core will be Fujitsu’s FUJITSU-MONAKA3 and MONAKA-X CPUs, fabricated with advanced 2-nanometer technology. These processors emphasize both high computational power and energy efficiency.

The architecture features optimized 3D packaging and ultra-low voltage circuits, allowing seamless integration with GPUs and other accelerators. This flexibility will enable the system to handle a diverse range of tasks—from complex scientific simulations to demanding AI workloads.

Collaboration and Strategic Vision

RIKEN has entrusted Fujitsu with the design and development of fugakunext’s overall system architecture, including computing nodes and CPUs. The basic design phase is set to complete by February 2026. Fujitsu’s leadership emphasizes delivering a system adaptable to evolving customer needs, leveraging their experience from the Fugaku project.

This project aligns with Japan’s broader “AI for Science” strategy, which combines artificial intelligence, simulation, and real-time data to accelerate discovery. The High-Performance Computing Infrastructure (HPCI) Steering Committee, under the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), highlights the growing demand for flexible, large-scale computing platforms. These platforms are essential for advancing research in generative AI and other data-intensive fields.

Looking Ahead

Fugakunext represents a significant leap in supercomputing capability, with the potential to reshape scientific research and AI development. As the project progresses, it will be vital to monitor how this technology influences global computational science and industry applications.

For researchers and professionals aiming to deepen their AI expertise, exploring current educational resources can be beneficial. Platforms like Complete AI Training’s latest AI courses offer practical learning paths that complement advancements in high-performance computing.