Japan's RIKEN institute, a global leader in scientific research, has announced a significant international collaboration with Fujitsu and NVIDIA to co-design FugakuNEXT, the successor to the acclaimed Fugaku supercomputer. According to details shared at the FugakuNEXT International Initiative Launch Ceremony in Tokyo, this initiative moves beyond mere speed, aiming to fundamentally reshape how high-performance computing (HPC) and artificial intelligence (AI) can address Japan's most pressing scientific and societal challenges.
The project is framed as a strategic investment in Japan's future, focusing on critical research priorities ranging from earth systems modeling and disaster resilience to advanced manufacturing and drug discovery. This effort is designed to integrate modern AI capabilities deeply into the nation's computational infrastructure, showcasing Japanese innovation for a global audience.
The announcement featured key figures including RIKEN President Makoto Gonokami and Satoshi Matsuoka, director of the RIKEN Center for Computational Science. Fujitsu Chief Technology Officer Vivek Mahajan and NVIDIA's Vice President of Hyperscale and High-Performance Computing, Ian Buck, also spoke, emphasizing the collaborative design approach and the platform's long-term potential.
This move follows NVIDIA CEO Jensen Huang's call last year for Japan to leverage the latest AI technologies to build its own sovereign AI infrastructure. FugakuNEXT directly responds to this, integrating NVIDIA’s comprehensive software stack, including CUDA-X libraries like cuQuantum for quantum simulation, RAPIDS for data science, TensorRT for high-performance inference, and NeMo for large language model development, alongside other domain-specific software development kits.
The Architecture of a New Scientific Era
FugakuNEXT is envisioned as a hybrid AI-HPC system, engineered to seamlessly combine traditional simulation workloads with advanced AI processing. At its core will be FUJITSU-MONAKA-X CPUs, designed to pair with NVIDIA technologies via NVLink Fusion. This new silicon innovation promises high-bandwidth connections between Fujitsu’s processing units and NVIDIA’s architecture, enabling unprecedented data flow and computational synergy. The system's design prioritizes not only raw speed and scale but also energy efficiency, a critical factor for next-generation supercomputing.
The applications targeted by FugakuNEXT are diverse and impactful. In scientific research, it will accelerate simulations through surrogate models and physics-informed neural networks, potentially automating hypothesis generation and code creation. For manufacturing, AI will learn from simulations to generate efficient and aesthetically pleasing designs at an accelerated pace. Crucially, in earth systems modeling, the supercomputer will enhance disaster preparedness and prediction capabilities for events like earthquakes and severe weather.
Collaboration extends beyond hardware, with RIKEN, Fujitsu, and NVIDIA jointly developing software tools for mixed-precision computing, continuous benchmarking, and performance optimization. This holistic approach underscores the project's ambition to create a fully integrated ecosystem for advanced research.
Backed by Japan’s MEXT (Ministry of Education, Culture, Sports, Science and Technology), FugakuNEXT will serve a broad spectrum of users, including universities, government agencies, and industry partners nationwide. It represents a strategic investment in sovereign infrastructure, fostering global collaboration while reaffirming Japan's commitment to leadership in scientific computing for the coming decade.
