- Jensen Huang reaffirmed Nvidia’s starring position within the AI business throughout a keynote deal with at Nvidia’s annual GTC convention on Tuesday. By means of its new open-source software program, Huang confirmed how Nvidia can ramp up DeepSeek R1’s effectivity 30-fold. But, whereas he spoke, Nvidia’s inventory worth dropped greater than 3%—after the corporate introduced its GPU timelines.
Clad in his signature black leather-based, Nvidia CEO Jensen Huang took heart stage at Nvidia GTC on Tuesday, defending the chip maker’s dominance within the business and touting the influence it may have on DeepSeek.
The occasion drew greater than 25,000 folks to the SAP Heart’s Nationwide Hockey League area, and Huang opened the keynote by launching t-shirts into the gang and coronating this 12 months’s GTC the “Super Bowl of AI.”
“The only difference is everybody wins at this Super Bowl, everybody’s a winner,” he joked. And just like the Tremendous Bowl, there have been GTC watch events and packed crowds to get a glimpse of Huang on stage.
Together with his deal with, Huang sought to dispel any uneasiness round AI funding, and mentioned dialogue about decrease spending doesn’t concern Nvidia. In January, apprehension engulfed the chip maker after it misplaced $589 billion in market cap in a single day after Chinese language AI reasoning mannequin Deepseek R1 claimed to function at a fraction of the price.
Whereas giant language fashions supply foundational information, reasoning fashions supply extra advanced, analytical responses. Utilizing the corporate’s new open supply software program Nvidia Dynamo, Huang mentioned the tech big’s Blackwell chips will be capable to make DeepSeek R1 30 instances sooner. He then performed a video demonstrating for the gang the way it could possibly be finished.
“Dynamo can capture that benefit and deliver 30 times more performance in the same number of GPUs in the same architecture for reasoning models like DeepSeek,” mentioned Ian Buck, vp and normal supervisor of Nvidia’s hyperscale and HPC computing enterprise.
From there, Huang’s keynote coated every little thing from the chip maker’s plans to roll out its latest chips— Blackwell Extremely later this 12 months, Vera Rubin in 2026, and Feynman in 2027.
“We have an annual rhythm of roadmaps that has been laid out for you,” Huang mentioned.
Whereas Nvidia’s introduced its strategic runway for years to come back, that wasn’t sufficient to cease the inventory’s slide. The chip maker’s share worth tumbled 3.4% Tuesday.
This story was initially featured on Fortune.com
