Nvidia's Jensen Huang Discusses DeepSeek's New AI Model

Nvidia CEO Jensen Huang recently spoke with CNBC's Jim Cramer during the annual GTC conference, revealing insights about DeepSeek's groundbreaking artificial intelligence model. Huang highlighted that the new R1 model from the Chinese startup, DeepSeek, is set to make waves in the AI industry, though its computational requirements are much higher than previously anticipated.

Describing the R1 model as "fantastic," Huang pointed out that this is the first open-sourced reasoning model. He elaborated on the model's capability to tackle problems by breaking them down into smaller steps, generating multiple potential answers, and then verifying the accuracy of these answers.

Huang stated, "This reasoning AI consumes 100 times more compute than a non-reasoning AI. This was exactly the opposite conclusion that everybody had." Such revelations underscore the surprising demands of advanced AI models compared to their predecessors.

The introduction of DeepSeek's model earlier this year inadvertently led to a significant sell-off in AI-related stocks. Investors became wary, fearing that this new model could rival existing top competitors while utilizing less energy and capital. Consequently, Nvidia experienced a notable decline, dropping 17% in one day and losing approximately $600 billion in market value, marking the largest single-day drop for any U.S. company.

During the interview, Huang also shared updates from Nvidia's conference, including advancements in AI infrastructure aimed at robotics and enterprise applications. He mentioned key partnerships with major companies like Dell, HPE, Accenture, ServiceNow, and CrowdStrike to bolster these efforts.

Reflecting on the evolving landscape of AI, Huang noted the shifting focus from generative AI to reasoning models. He also projected that the computing capital expenditures globally would likely reach 1 trillion dollars by the decade's end, emphasizing that a substantial portion of these funds will be directed towards AI innovations.

Huang concluded by stating, "So, our opportunity as a percentage of a trillion dollars by the end of this decade is quite large. We've got a lot of infrastructure to build."

Nvidia, DeepSeek, AI