Real Cloud Infrastructure for Real AI Workloads: Training and Inference at Production Scale

What you’ll learn in this on-demand session

In this session, CoreWeave detailed how we optimized every layer of the AI stack—from infrastructure to orchestration to observability—to efficiently run large-scale training and inference workloads. We also examined the architectural breakthroughs that enabled rack-scale systems to operate with ultra-low latency and high reliability.

These were the essential cloud components that powered the next generation of agentic AI. The question was: How did your infrastructure stack up?

We cover:

• How infrastructure requirements changed when scaling to trillion-parameter and mixture-of-experts models
• How full-stack optimization across infrastructure, orchestration, and observability improved performance and efficiency
• Architectural innovations that enabled ultra-low latency, rack-scale AI systems
• Best practices for running production-grade AI workloads, including agentic AI systems