Core42 Achieves Top-20 Global Ranking for Maximus-01 AI Supercomputer
Through its collaboration with AMD, Core42’s Buffalo deployment features more than 9,000 AMD Instinct MI300X GPUs configured for large-scale AI training and inference.
Core42, a G42 company specializing in sovereign cloud and AI infrastructure, has announced that its Maximus-01 supercomputer has been ranked No. 20 globally on the TOP500 List, the industry’s leading benchmark for publicly known supercomputing systems. The ranking recognizes the performance of the AMD Instinct™ MI300X GPU-based Maximus cluster, operated at Core42’s facility in Buffalo, New York, ahead of Supercomputing 2025, the premier global conference for high-performance computing (HPC).
The TOP500 List measures supercomputing performance using the High Performance Linpack (HPL) benchmark. Achieving a top-20 position reflects exceptional engineering across compute, memory, storage, and networking, and validates Maximus-01’s ability to support large-scale AI and HPC workloads under sustained production conditions.
Through its collaboration with AMD, Core42’s Buffalo deployment features more than 9,000 AMD Instinct MI300X GPUs configured for large-scale AI training and inference. Delivered via the Core42 AI Cloud, the cluster provides organizations with access to advanced GPU capacity supported by unified orchestration, enterprise-grade security, and comprehensive data management.
The Core42 AI Cloud infrastructure is purpose-built for HPC workloads optimized for AI. Its compute island operates as a self-contained environment with dedicated high-performance storage and a backbone network delivering exceptional speed and reliability. The heterogeneous AI Cloud architecture also integrates Nvidia and Cerebras accelerators and is powered by Broadcom high-bandwidth networking chips and Arista Networks’ Ethernet switches and EOS software, which optimize traffic flow and data routing.
The Maximus-01 cluster offers organizations enterprise-grade AI infrastructure capable of handling advanced GPU workloads with regulatory compliance and low-latency performance. Its placement on the TOP500 List marks an important milestone for Core42 and reinforces the company’s capacity to operate global-scale, high-performance computing environments. The system achieves a Linpack performance (Rmax) of 114.50 petaflops and a theoretical peak (Rpeak) of 251.15 petaflops.
The Buffalo cluster operates within TeraWulf’s Lake Mariner facility as part of G42’s strategic partnership with the company. The site provides power resiliency, cooling efficiency, and operational scale for large GPU deployments, combining Core42’s AI infrastructure with TeraWulf’s zero-carbon, energy-efficient compute environment. The facility supports scalable AMD Instinct MI300X GPU capacity for advanced AI training and inference workloads while maintaining enterprise-grade performance, reliability, and environmental standards.
The achievement highlights the power of combining cutting-edge GPUs with world-class AI infrastructure, reinforcing Core42’s mission to provide scalable, high-performance computing environments that accelerate AI and HPC innovation globally.
