The Nvidia (NASDAQ: NVDA) GPU Technology Conference (GTC) has become a big event for the artificial intelligence developer community in recent years, and when CEO Jensen Huang gave his keynote speech at this year’s event last week, he had some noteworthy thoughts to share about where AI is going and news about Nvidia’s place in its future.
If you’re a Nvidia shareholder or are considering becoming one, here are three important updates you need to know about.
Nvidia only recently rolled out its newest and most powerful graphics processing units (GPU), built on its latest Blackwell architecture, and demand has been intense.
Nearly every developer and platform is looking for ways to leverage AI, and Nvidia has the most effective products to support those efforts. In its fiscal 2025’s fourth quarter, which ended Jan. 26, revenue swelled by 78% year over year. Huang noted that Nvidia already had already booked billions of dollars in sales of Blackwell in its first quarter on the market. Tech giants with the budgets to build massive AI data centers want the best, and none of them wants to fall behind their peers.
But Nvidia is already developing its next iteration of powerful chips: the Rubin architecture. These will be 14 times more powerful than Blackwell, and as AI’s needs evolve and expand, it’s a no-brainer bet that the companies developing and supporting the software will want the most powerful chips they can buy. Rubin is expected to launch late next year.
Agentic AI is one of the next waves in the artificial intelligence space. These specialized tools will be able to act as “agents” on a user’s behalf, autonomously taking care of such tasks as booking flights or writing emails after being prompted with simple requests.
Because that will mean that such systems will need to manage far more steps, Huang said he thinks that agentic AI is going to need 100 times more power than current AI tools need. He thinks that those who believe that the advent of tools like DeepSeek — which was apparently much cheaper to develop than previous large language models — are going to undermine Nvidia’s sales are getting it wrong, because the sheer power requirements of agentic AI will mean a greater need for chips like Nvidia’s.
While some peers have similar offerings on the market, Nvidia dominates in terms of production and sales in cutting-edge GPUs and AI accelerators, and it’s constantly improving its chips to retain that dominance.
