Nvidia discusses $2,000-per-GPU annual software fee for running agents on AI infrastructure

Huang discussed the pricing strategy at GTC Berlin on October 22, outlining Nvidia's broader push to build an AI infrastructure business spanning chips, software, and AI-agent security. The proposed fee structure represents a significant shift in the company's revenue model, emphasizing software licensing as a major component of future earnings.

Future Revenue Stream

Under a scenario outlined by FourWeekMBA involving five million paid licenses, this pricing model could generate approximately $50 billion over five years for Nvidia. The company already charges more than twice the proposed amount for its AI Enterprise software platform.

Market Expansion

This move comes as Nvidia continues to dominate the AI chip market while expanding its software offerings. The company's strategy aims to create a comprehensive ecosystem where customers not only purchase hardware but also commit to ongoing software licensing, potentially increasing customer lifetime value.