Corey Sanders, SVP of Product at CoreWeave, discusses why AI infrastructure fundamentally differs from traditional cloud computing. The conversation covers AI-native infrastructure design, the shift from training to inference-heavy workloads, the rise of agentic systems, GPU optimization strategies, and why the future of software will be built around AI-first experiences rather than conventional web and app architectures.
This summary was generated from show notes and public descriptions, not from a full transcript review. Details may contain inaccuracies.
Novel
•
Agentic Systems Require New Infrastructure Primitives
AI agents that make decisions, call tools, and iterate require infrastructure patterns that traditional cloud platforms do not optimize for.
AI Infrastructure Is Fundamentally Different From Traditional Cloud
AI workloads—training, inference, and agentic systems—require purpose-built infrastructure that differs radically from general-purpose cloud computing.
The Shift From Training to Inference-Heavy Workloads
As AI moves into production, inference becomes the dominant compute workload, fundamentally changing how infrastructure must be designed and optimized.
GPU Performance Optimization Becomes a Differentiator
Extracting maximum throughput from GPU hardware through driver tuning, memory management, and network optimization is a specialized skill that infrastructure providers must master.
Software Will Be Built Around AI-First Experiences, Not Websites
The next generation of software will not be web apps with AI features bolted on, but applications designed from the ground up with AI as the primary interface.