{"slug": "arm-powers-the-worlds-most-advanced-and-energy-efficient-supercomputers", "title": "Arm powers the world’s most advanced and energy-efficient supercomputers", "summary": "Arm architecture powers the world's most advanced and energy-efficient supercomputers, including the new TOP500 #1 system LineShine, which exceeded two exaflops, and Europe's first exascale system JUPITER. Arm Neoverse-based NVIDIA Grace systems claimed the top four positions on the Green500, demonstrating that organizations no longer need to choose between performance and efficiency.", "body_md": "# Arm powers the world’s most advanced and energy-efficient supercomputers\n\n*This blog was written by Koray Ozkal, Senior Product Marketing Manager, Cloud AI and Conrad Hillariet, Principal HPC Engineer, Ecosystem & Partner Engagement*\n\nHPC has always pushed the boundaries of what is possible, from climate modeling and scientific research to drug discovery and engineering simulation. As HPC, AI, and cloud infrastructure converge, these increasingly AI-driven workloads require computing architectures that can scale efficiently while delivering the performance, power efficiency, and cost efficiency organizations need. Arm is helping meet that challenge.\n\n## From Exascale to AI: Arm powers the future of supercomputing\n\nNowhere is that more evident than in the latest [TOP500](https://top500.org/) and [Green500](https://top500.org/lists/green500/green500-june-2026/) rankings, announced at ISC 2026 on June 23. The latest lists reinforce the growing momentum of Arm architecture across the world’s highest-performing and most energy-efficient supercomputers. TOP500’s new #1 system, LineShine, became the first supercomputer to exceed two exaflops of sustained double-precision performance. It features a CPU-only design compatible with Arm architecture Armv9. At the same time, from Europe’s first exascale system to the world’s most energy-efficient supercomputers, Arm Neoverse-based NVIDIA Grace systems continue to lead the Green500, claiming the top four positions.\n\nArm Neoverse technology is enabling organizations to achieve breakthrough performance while improving power efficiency and total cost of ownership. One of the most significant milestones is JUPITER, [Europe’s first exascale supercomputer](https://blogs.nvidia.com/blog/jupiter-exascale-supercomputer-live/) and currently Europe’s fastest supercomputer. Built on Bull’s [BullSequana XH3000](https://www.bull.com/en/products/hpc/bullsequana-xh3000) liquid-cooled architecture, JUPITER is powered by nearly 24,000 NVIDIA GH200 Grace Hopper Superchips containing more than 1.7 million Arm Neoverse cores. Designed to tackle some of the world’s most complex scientific and industrial challenges, JUPITER demonstrates how Arm’s performance and efficiency can scale to exascale computing.\n\nWe are pleased that Europe’s first exascale supercomputer, JUPITER, is powered with Arm architecture. The NVIDIA Grace CPUs and SiPearl Rhea processors deliver the performance, scalability and energy efficiency needed for advanced HPC and AI workloads. We see Arm technologies playing a central role in the future of AI and HPC, and we look forward to continued collaboration to drive scientific breakthroughs.”\n\n– Thomas Lippert, Director of the Jülich Supercomputing Centre.\n\nEnergy efficiency is equally critical in modern supercomputing. As power availability becomes one of the primary constraints on future HPC growth, energy efficient compute architectures are becoming essential. The latest Green500 rankings reinforce Arm’s leadership in this area.\n\nKairos, powered by Arm Neoverse-based NVIDIA GH200 Grace Hopper Superchips, currently ranks as the world’s most energy-efficient supercomputer. Kairos is joined by fellow Arm Neoverse-based NVIDIA systems ROMEO, Levante, the top three most energy-efficient supercomputers in the world. Together, these systems demonstrate that organizations no longer need to choose between performance and efficiency.\n\nBy leveraging Arm Neoverse-based CPUs and Bull’s Direct Liquid Cooling technology, we overcame strict power envelope constraints to secure the top three spots on the Green500 for four consecutive editions with KAIROS, ROMEO and Levante systems while also powering Europe’s first exascale system, JUPITER. Our Arm-based CPU portfolio, integrating both NVIDIA and SiPearl processors within our BullSequana XH platforms, proves that extreme HPC-AI performance and world-class energy efficiency can seamlessly coexist.\n\n– Bruno Lecointe, SVP HPC AI & Quantum Business line, Bull.\n\nPurpose-built for AI research, [Isambard-AI](https://blogs.nvidia.com/blog/isambard-ai/) is the United Kingdom’s most powerful AI supercomputer. Powered by NVIDIA Grace Hopper Superchips, it is helping researchers accelerate breakthroughs in areas ranging from automated drug discovery to climate science.\n\nThe University of Bristol has been working with Arm since 2009, leading to Bristol joining the Mont Blanc 2 European project. In 2018, Bristol built and deployed Isambard 1 based on Marvell ThunderX2 – this was the world’s first production supercomputer using Arm CPUs. Bristol’s Isambard series of supercomputers has continued this pioneering work ever since.\n\nIn 2020, Isambard 2 adopted Fujitsu’s A64fx and in 2024, Isambard 3 became one of the first systems in the world to use NVIDIA’s new Grace Arm CPU. In 2025, the UK’s most powerful supercomputer, Isambard-AI, launched with 5,448 NVIDIA Grace Hopper GPUs, underpinned by Arm-based CPU architecture, AArch64.\n\nHaving worked with Arm technologies for more than 15 years, we have seen first-hand how the architecture has evolved to support increasingly demanding scientific computing workloads. We are excited to work further with Arm as they advance the capabilities of AI-driven research and enable world-changing scientific discoveries.\n\n– Simon McIntosh-Smith, Director of the Bristol Centre for Supercomputing.\n\n## From Fugaku to FugakuNEXT: Japan’s fastest supercomputing\n\nWhen Fujitsu’s Arm-compatible A64FX processor powered Fugaku to become the world’s fastest supercomputer, it demonstrated that Arm Arch v8 architecture could handle the most demanding computational workloads on the planet. Today, Fugaku remains the fastest supercomputer in Japan, the second fastest in the Asia-Pacific region and one of the world’s leading HPC systems, built on 158,976 Fujitsu A64FX processors.\n\nThat momentum will continue with Fujitsu’s [next-generation roadmap](https://global.fujitsu/-/media/Project/Fujitsu/Fujitsu-HQ/technology/research/events/sc25/SC25-FUJITSU-MONAKA.pdf?rev=138be9d7353849f3982e64617b347e20&hash=10F0FFEC2C6D5B703416334F3889D409). [FUJITSU-MONAKA](https://global.fujitsu/en-global/technology/research/fujitsu-monaka), expected to reach general availability in 2027, represents another major milestone in Arm’s role in HPC. Fujitsu has already announced its successor, FUJITSU-MONAKA-X, which is planned to power FugakuNEXT and continue the company’s long tradition of supercomputing innovation on Arm.\n\nA key advantage behind Arm’s HPC momentum is its software ecosystem. From open-source HPC applications and AI frameworks to tooling, developers can take advantage of a mature software environment that already supports Arm. This allows organizations to deploy applications across Arm-compatible systems with minimal effort while benefiting from a common architecture spanning cloud, AI and HPC infrastructure. Arm Neoverse-based NVIDIA Grace CPUs are one example of this software ecosystem advantage. The majority of [software that runs on Grace benefits from the broader Arm software ecosystem](https://developer.nvidia.com/blog/nvidia-grace-cpu-integrates-with-the-arm-software-ecosystem/), while NVIDIA has also optimized its accelerated computing stack for Arm. This creates a powerful foundation for innovation, allowing software advancements developed for Grace to benefit a growing range of Arm Neoverse-based platforms.\n\nThe Arm software ecosystem continues to expand with the introduction of [Arm Performix](https://developer.arm.com/servers-and-cloud-computing/arm-performix), a new performance analysis and optimization tool designed to help developers extract maximum efficiency from Arm-based systems. Performix provides actionable insights into application behavior, enabling developers to identify performance bottlenecks, optimize code and make better use of modern Arm architectures. By simplifying performance tuning across HPC, AI and cloud workloads, Performix complements the existing ecosystem of compilers, libraries and development tools, helping organizations accelerate application deployment while achieving higher levels of performance and energy efficiency on Arm-powered platforms.\n\n## Arm Neoverse outperforms legacy x86 architectures\n\nThe performance benefits of Arm technologies are quite significant. Across key HPC workloads including computational fluid dynamics, weather modeling and molecular dynamics, Arm Neoverse-powered NVIDIA Grace systems have demonstrated [up to 3x higher performance per watt compared to legacy x86-based platforms](https://developer.nvidia.com/blog/nvidia-grace-cpu-delivers-high-bandwidth-and-efficiency-for-modern-data-centers/). This enables organizations to increase computational throughput while reducing energy consumption and infrastructure costs.\n\nBeyond traditional scientific computing, Arm is increasingly becoming the platform of choice for the world’s leading compute builders. From AWS Graviton, Microsoft Cobalt and Google Axion to NVIDIA Grace and NVIDIA Vera platforms, Arm-based CPUs are powering the next generation of cloud, AI and HPC infrastructure. Arm technology is also increasingly serving as the orchestration layer for rack-scale and system-level designs, including NVIDIA’s rack level [Grace Blackwell](https://www.nvidia.com/en-us/data-center/gb300-nvl72/?ncid=no-ncid) and [Vera Rubin](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/?ncid=no-ncid) systems, [AWS Trainium3 UltraServers](https://newsroom.arm.com/blog/arm-converged-ai-data-center-aws-graviton5) and [Google Cloud TPU SuperPods](https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/eighth-generation-tpu-agentic-era/).\n\n## Compute for the agentic era\n\nThis evolution continues with the recently [announced Arm AGI CPU](https://newsroom.arm.com/blog/introducing-arm-agi-cpu?utm_source=google&utm_medium=cpc&utm_content=text_txt_na_brand-event&utm_campaign=mk30_brand-paid_everywhere_keyword_traffic_brand-sitelink&utm_term=arm%20cpus&gad_source=1&gad_campaignid=23672950280&gbraid=0AAAAADLDCY-Vnf9JLpvVgOJLi1itcsyMS&gclid=Cj0KCQjw3K7RBhDJARIsAKRtP5RrUJJXRbH63QmJAMBBPW0UGFT36LWgoBz45ycnYW1462YMDFnstv0aAopHEALw_wcB), Arm’s first production silicon designed specifically for the gigawatt-scale datacenters powering agentic AI. Acting as a central head node, the [Arm AGI CPU](https://www.arm.com/products/cloud-datacenter/arm-agi-cpu) orchestrates or executes continuous tasks and workloads, manages accelerators and coordinates Agentic workflows across increasingly complex computing environments.\n\n## Learn more about Arm HPC\n\nAs supercomputing, AI and cloud infrastructure continue to converge, one thing is becoming clear: the future of high-performance computing will require platforms that deliver exceptional performance, efficiency, scalability and software compatibility. Arm Neoverse is helping enable that future today.\n\nTo learn more about Arm’s work in high-performance computing and the growing Arm HPC ecosystem, visit: [https://www.arm.com/markets/cloud-ai/high-performance-computing](https://www.arm.com/markets/cloud-ai/high-performance-computing)\n\nAny re-use permitted for informational and non-commercial or personal use only.", "url": "https://wpnews.pro/news/arm-powers-the-worlds-most-advanced-and-energy-efficient-supercomputers", "canonical_source": "https://newsroom.arm.com/blog/arm-powers-energy-efficient-hpc-supercomputers", "published_at": "2026-07-15 16:00:00+00:00", "updated_at": "2026-07-15 16:14:27.587034+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-infrastructure", "ai-chips", "ai-products"], "entities": ["Arm", "NVIDIA", "JUPITER", "LineShine", "Kairos", "Bull", "SiPearl", "Green500"], "alternates": {"html": "https://wpnews.pro/news/arm-powers-the-worlds-most-advanced-and-energy-efficient-supercomputers", "markdown": "https://wpnews.pro/news/arm-powers-the-worlds-most-advanced-and-energy-efficient-supercomputers.md", "text": "https://wpnews.pro/news/arm-powers-the-worlds-most-advanced-and-energy-efficient-supercomputers.txt", "jsonld": "https://wpnews.pro/news/arm-powers-the-worlds-most-advanced-and-energy-efficient-supercomputers.jsonld"}}