From Chip to Megawatt: System Presented for Large Clusters Training AI Models

The race for artificial intelligence dominance is extending beyond processing power, with memory, networks, software, energy, and the ability to integrate thousands of accelerators into reliable industrial systems becoming crucial. This shift is highlighted by a partnership between AMD and Schneider Electric, aimed at building the necessary physical infrastructure for Europe's goal of becoming an "AI continent."

On July 23, the two companies unveiled the first jointly developed and validated infrastructure blueprint for AMD Helios. This system is specifically designed for large clusters that train and run artificial intelligence models. The blueprint pre-integrates computing equipment with power supply, cooling, physical layout, and data center management software, streamlining the deployment of AI infrastructure.

This collaboration underscores the complex challenges involved in translating strategic ambitions and public investments into tangible AI infrastructure. AMD's potential role in facilitating this transition is significant, as the AMD Helios system offers a comprehensive solution for managing the demanding requirements of large-scale AI operations. Further details are available on Schneider Electric's website.