NVIDIA CEO Unveils Bold AI Vision: Preparing for a 100-Fold Surge in Computational Demand

NVIDIA CEO Jensen Huang drew considerable attention at the recent GTC 2025 conference with his impromptu keynote address, highlighting significant advancements in AI. One of the main points of his presentation concerned the demand for computational power, which he suggested has been underestimated by the market. Huang pointed specifically to models like DeepSeek, which he said require significantly more computational support than previously anticipated.

Huang described the evolution of AI, emphasizing that the emergence of "Agentic AI" signifies a shift toward autonomous systems capable of reasoning and decision-making within complex environments. This evolution alters traditional computational models, increasing the demand for powerful AI systems capable of generating vast quantities of tokens. According to Huang, the current need for computation exceeds projections from just a year ago, with increases potentially reaching 100 times the previous estimates.

The remarkable demand is attributed to advances in AI reasoning techniques, like "Chain of Thought." These techniques decompose problems, allowing AI to generate more intricate and extensive solutions. Consequently, the necessity for a higher volume of computational tokens, and increased processing speeds, leads to this unprecedented leap in computational requirements.

NVIDIA is addressing the burgeoning demands of AI by maintaining a consistent annual release schedule for its new chips. Huang announced an ambitious roadmap extending to 2028, featuring successive launches such as the Blackwell Ultra and the Rubin platforms, which promise to leverage new-generation technologies to meet these demands effectively. The planned advances in AI hardware are intended to support NVIDIA's overarching goal of fostering fully functional AI factories, designed to handle the intense computational loads of the future.

In conclusion, Huang made it clear that understanding and planning for the future of AI is tantamount to laying the groundwork for these expansive 'AI factories.' This necessitates diligent planning on all fronts—from infrastructure and power to team deployment. By laying out a clear path for the next few years, NVIDIA underscores its commitment to leading in the field of AI computation, ensuring its developments align with the accelerating pace of AI technology.