Courtesy of DDN
- Investors misinterpreted DeepSeek's AI advancements, Nvidia CEO J...
NVIDIA CEO: Investors Misinterpreted DeepSeek's AI Advancements
NVIDIA Stock Sold Off on Misperceptions
NVIDIA CEO Jensen Huang believes investors reacted incorrectly to DeepSeek's AI breakthroughs during a virtual event on Thursday.
DeepSeek's AI Innovations: A Challenge or Opportunity?
DeepSeek, a Chinese AI firm, released a competitive open-source reasoning model in January, sparking concerns about the need for vast AI infrastructure spending. However, Huang highlighted that post-training methods, where AI models make conclusions after training, require significant computing power.
The Importance of Post-Training Methods
As post-training methods evolve, the demand for the advanced computing power provided by NVIDIA chips will increase. Huang emphasizes that pre-training, while crucial, is only one aspect of AI intelligence; post-training plays a vital role in problem-solving.
DeepSeek's Impact on the AI Landscape
Huang applauds DeepSeek's innovations, describing the global excitement surrounding its open-source model as "incredible." The company's advancements have sparked innovation and discussion throughout the AI industry.
NVIDIA's Response to Concerns
NVIDIA has addressed the market reactions through written statements, while Huang made his first public comments on the topic at Thursday's event. He has consistently emphasized that model scaling remains relevant but has shifted from training to inference.
Post-Training Methods Drive AI Improvements
Huang stressed that post-training methods are intensive and that models will continue to improve with the development of new reasoning methods. These comments may offer insights into NVIDIA's upcoming earnings call.
DeepSeek as an Earnings Call Topic
DeepSeek has become a popular discussion point on earnings calls for tech companies, including Airbnb and Palantir. NVIDIA rival AMD's CEO, Lisa Su, acknowledged DeepSeek's impact on AI innovation and adoption.
