Nvidia Unveils New AI Products to Boost DeepSeek’s Model Capabilities
Nvidia is rolling out a suite of new products designed to significantly enhance the performance of AI models, specifically targeting improvements for models like DeepSeek’s R1. The announcement was made during the company’s annual GTC conference in San Jose, California, where CEO Jensen Huang highlighted the potential of these new offerings.
During his keynote, Huang emphasized how Nvidia’s latest products can dramatically boost the reasoning capabilities of large models. A key component of this strategy is Dynamo, Nvidia’s new open-source inference software. Huang described Dynamo as the “operating system of an AI factory.” According to Nvidia, Dynamo is available on Github and can deliver up to 30 times greater performance on existing GPUs and architectures for AI reasoning tasks.
The announcement comes at a time when Nvidia’s dominant position in the AI sector faces increased scrutiny. DeepSeek, a Hangzhou-based company, unveiled low-cost, high-performance models in January, raising questions about the need for expensive AI chips. Nvidia, which faces US export restrictions preventing the sale of advanced chips to China, has been working to address concerns about the “scaling law,” which suggests that AI systems improve with more resources.
Nvidia’s shares saw a slight dip, falling 3.4% to US$115.4 on Tuesday, although this was still above the price on January 27 when the stock lost 17% in a single day.
