Supermicro Expands AI-RAN and Sovereign AI Support

Sovereign AI enables nations to control their own data and AI infrastructure, ensuring that sensitive information remains within their borders and complies with local regulations. This addresses growing concerns about data privacy and national security in an era of globally accessible AI platforms. Supermicro is partnering with companies like Telenor in Norway and SK Telecom in South Korea to build these in-country "AI factories." SK Telecom's "Haein Cluster" is a prime example, utilizing over 1,000 Supermicro AI servers equipped with NVIDIA Blackwell GPUs to create one of South Korea's most powerful AI clusters. Similarly, Telenor's AI Factory provides a sovereign AI cloud for Norwegian enterprises, ensuring data remains within national borders. These initiatives leverage Supermicro's modular Data Center Building Block Solutions to accelerate deployment. The move also heavily supports AI-RAN (Artificial Intelligence-Radio Access Networks), which integrates AI to optimize wireless network performance and efficiency. This allows for intelligent traffic management, enhanced spectrum use, and better resource allocation, turning the RAN from a cost center into a potential revenue source by hosting new AI services at the network edge. To power these solutions, Supermicro is using the latest silicon, including systems based on the NVIDIA GH200 Grace Hopper Superchip. The GH200 combines a 72-core Grace CPU and a Hopper GPU with a 900 GB/s coherent interface, which is 7 times faster than PCIe Gen5, providing direct GPU access to up to 624GB of combined fast memory. On the Intel side, the architecture leverages Intel Xeon 6 processors, which feature Performance-cores (P-cores) and Efficient-cores (E-cores) to handle diverse workloads from AI to networking. The Xeon 6 includes built-in AI acceleration with Intel Advanced Matrix Extensions (AMX) to speed up inference and training tasks directly on the CPU. For telecom-specific deployments, Supermicro offers systems like the short-depth 1U ARS-111L-FR, designed for distributed RAN workloads with an NVIDIA Grace CPU. This aligns with NVIDIA's Aerial RAN Computer (ARC) design, demonstrating a collaborative effort with partners like Nokia to validate and deploy these AI-native network solutions.