Nvidia Lifts AI-Chip Outlook to $1 Trillion Through 2027 at GTC

Overview

• CEO Jensen Huang updated the revenue opportunity for Nvidia’s AI chips to at least $1 trillion through 2027, replacing an earlier ~$500 billion view for 2026.
• Nvidia will add the Groq 3 LPU to its catalog as a language-model inference co-processor, with Samsung set to manufacture the Groq-derived chip for systems arriving in the second half of 2026; the move follows a December licensing deal that brought many Groq engineers to Nvidia.
• The company is expanding into general-purpose computing with plans to sell computers built entirely from CPUs, complementing its accelerator-driven platforms.
• Next-generation Vera/Rubin-class processors are expected to appear in systems in the second half of 2026.
• Nvidia shares closed up 1.6% after an intraday gain of as much as 4.8%, as Huang underscored CUDA and the installed base as key advantages while competition from AMD, Intel and in‑house customer chips persists.