Why extreme co-design beats faster chips with Ian Finder at NVIDIA

Alex sits down with Ian Finder, Head of Data Center CPU at NVIDIA, to unpack the design philosophy behind the Vera CPU, NVIDIA's clean-sheet core built for a world where AI agents, not just humans, are the primary consumers of compute. They trace the lineage from Grace to Vera, dig into chiplet-based coherency fabrics, why deterministic tail latency matters as much as raw throughput, and how agentic workloads are pushing compute load back onto the CPU. They also get into SLA classes for different agent workloads, the parallel between agent state swapping and GPU KV cache pressure, and Ian's home-lab experiments running local open-weight models to orchestrate FPGA hardware simulation.