In this DCF Trends-Nomads at the Summit Podcast episode, Chris James, CEO of NoesisAI, delivers a sweeping, insight-rich overview of how different classes of AI models—from LLMs and RAG to vision AI and scientific workloads—are driving a new wave of infrastructure decisions across the data center landscape. With a sharp focus on the diverging needs of training vs. inference, James breaks down what it takes to support today’s AI—from GPU-intensive clusters with high-speed interconnects and liquid cooling to inference-optimized, edge-deployed accelerators. He also explores the rapidly shifting hardware ecosystem, including the rise of custom silicon, heterogeneous computing, and where the battle between NVIDIA, AMD, Intel, and hyperscaler-designed chips is headed. Whether you're designing for scalability, sustainability, or the bleeding edge, this conversation offers a field guide to the infrastructure behind intelligent computing.