DeepSeek, a Chinese startup, has made headlines with its groundbreaking R1 model, the first open-sourced reasoning AI model, which was unveiled recently. The R1 model, discussed extensively at Nvidia's annual GTC conference, is considered a significant development in the field of artificial intelligence. However, the announcement has also triggered a massive sell-off in AI stocks, as investors fear the model's potential to perform comparably to top competitors while requiring less energy and financial resources.
Nvidia's CEO Jensen Huang discussed the implications of DeepSeek's R1 model in an interview with CNBC's Jim Cramer during the conference. Huang highlighted that the R1 model requires 100 times more computation than non-reasoning AI models, a revelation that caught many in the industry by surprise.
"This reasoning AI consumes 100 times more compute than a non-reasoning AI." – Jensen Huang
The R1 model's consumption of computational resources has raised eyebrows, particularly because it challenges the assumption that reasoning models could operate efficiently with fewer demands. Huang underscored this point during the interview.
"It actually consumes 100 times more computing." – Jensen Huang
Despite the concerns about resource consumption, DeepSeek's R1 model is lauded for its potential impact on the industry. As an open-sourced model, it provides unprecedented access to reasoning AI capabilities, allowing for broader experimentation and innovation within the AI community. This move by DeepSeek positions them as a formidable player in the AI landscape.
The sell-off in AI stocks reflects investor anxiety over the R1 model's capabilities and its implications for existing players in the market. The fear is that DeepSeek's model could match or exceed the performance of leading AI technologies, thereby reshaping competitive dynamics and reducing the cost-effectiveness of current solutions.