Jensen Huang discusses the economics of inference, power delivery, and more at CES 2026 press Q&A session — 'You sell a chip one time, but when you build software, you maintain it forever'

Nvidia CEO Jensen Huang discussed various tech topics at CES 2026, emphasizing the importance of maintaining system productivity in large-scale AI deployments. Huang highlighted the economic impact of system downtime and the need for efficient maintenance strategies, such as Nvidia's Vera Rubin platform designed for serviceability. The focus on power delivery optimization in AI systems, particularly managing demand spikes, was a key theme, with Nvidia aiming to flatten power usage profiles to enhance efficiency. Huang also underscored the significance of inference economics, prioritizing flexibility in software and memory models to adapt to evolving AI workloads. Additionally, the rise of open models was addressed as a driver of increased inference demand, with Nvidia positioning itself to support a broader range of deployment scenarios.