Nvidia founder and CEO, Jensen Huang, expressed that there has been a misunderstanding in the market concerning the technological advancements made by DeepSeek and the potential negative impact on Nvidia’s business. During a pre-recorded interview with Alex Bouzari, CEO of DataDirect Networks, Huang described DeepSeek’s R1 open-source reasoning model as “incredibly exciting.”
Huang commented that the market’s reaction to R1 was, “Oh my gosh. AI is finished,’” suggesting that it negated the need for computing power. He argued that this perspective is completely incorrect. Huang believes that the introduction of R1 is beneficial for the AI market, as it will enhance the adoption of AI rather than diminish the need for computational resources such as those produced by Nvidia.
He further explained that the release of R1 draws attention to the potential for models to be more efficient than previously believed, thus accelerating AI adoption. Despite DeepSeek’s advances in pre-training AI models, Huang highlighted the continuous importance of post-training, which remains resource-intensive, stating that, “Reasoning is a fairly compute-intensive part of it.”
Nvidia has declined to provide additional commentary on the matter. Huang’s remarks follow the release of DeepSeek’s R1 model, which significantly affected the AI market and had a notable impact on Nvidia. The company’s stock price fell by 16.9% in one day following DeepSeek’s announcement.
Nvidia’s stock closed at $142.62 per share on January 24, but by January 27, it had dropped to $118.52 per share, effectively wiping $600 billion from Nvidia’s market cap in just three days. However, the company’s stock has almost fully recovered, opening at $140 per share on Friday, almost regaining the lost value within a month. Nvidia is expected to address the market reaction in its Q4 earnings report on February 26.
Meanwhile, DeepSeek announced plans on Thursday to open source five code repositories during an “open source week” event in the coming week.
