At COMPUTEX 2026, Nvidia CEO Jensen Huang signaled a strategic pivot away from GPU-centric infrastructure, announcing that CPU will become the company's next multibillion-dollar business growth engine. Nvidia unveiled Vera, its first self-designed data center CPU featuring 88 custom Olympus cores and spatial multithreading technology specifically engineered for AI agents and long-context management. Huang projects the CPU market opportunity at $200 billion.
The shift reflects AI workload migration from model training to large-scale inference and agent deployment. Research shows CPUs handle 50% to 90% of total latency in agent workflows, requiring reconfiguration from traditional 1:8 CPU-to-GPU ratios toward 1:4 or even 1:1 configurations. AMD CEO Su Zifeng echoed the trend, predicting data center CPU market compound annual growth exceeding 35% over five years, while Intel reported Q1 2026 net profit surging 156% year-over-year. Server CPU lead times now extend 6 months for Intel and 8–10 weeks for AMD, reflecting supply constraints.