NVIDIA GTC 25: ‘Everybody Wins’ at the Super Bowl of Artificial Intelligence

Huang’s was a different kind of keynote address than what you might normally see at a major technology conference. While most CEOs deliver prepared remarks before yielding the floor to other executives and partners, Huang held the stage alone for more than two hours without the assistance of notes or a teleprompter.

It’s part of his mystique, and prior to his taking the stage, a series of other technology giants and investors paid him homage in a pre-event show that was aired to attendees, including Michael Dell, founder of Dell Technologies, and Bill McDermott, CEO of ServiceNow.

“He’s made a bunch of bold, incredibly insightful decisions,” Dell said. “He saw the importance of all of this before almost anyone else, bet big on it and led with relentless precision and vision.”

FIND OUT: Can artificial intelligence agents really ease workloads for large enterprises?

The Era of Agentic AI Is Underway, and Physical AI Is Next

When he took the stage, Huang marveled at how far AI has advanced in the just the past few years. It began barely more than a decade ago with “perceptive” AI, which included early speech recognition and computer vision. Frustrating customer service interactions with bots were part of this generation of AI. Then came generative AI, which opened the world’s eyes to the power of the technology.

Today, the world is in the era of agentic AI, which Huang called a “major breakthrough” and “a fundamental advance” in the technology. “Agentic AI basically means you have an AI that has agency,” he explained. “It can understand the context of circumstances, it can reason about how to solve a problem and it can plan action. It can use tools. It can go to a website, scan the website, maybe play a video, learn from the website and then come back and use the knowledge to do its job.”

“The amount of computation we need as a result of agentic AI, as a result of reasoning, is easily 100 times more than we thought we needed this time last year," he said. For this reason, Huang said, a data center building boom is underway. He predicted that spending on new data centers would soon reach $1 trillion globally.

The era of physical AI, which includes technologies such as autonomous vehicles and robots that perform a wide array of tasks, is also underway but at an earlier stage, he said. It was on the physical AI front that Huang delivered NVIDIA’s first big announcement: An expanded partnership with General Motors, in which NVIDIA will help the carmaker deliver Level 3 autonomous driving capabilities. These will require a driver’s presence, but the AI does most of the driving, in certain models.

As part of the partnership, GM will also use NVIDIA technology to create digital twins of its factories. Digital twins are virtual replicas of physical assets, allowing manufacturers to optimize processes and fix inefficiencies. “AI not only optimizes manufacturing processes and accelerates virtual testing but also helps us build smarter vehicles while empowering our workforce to focus on craftsmanship,” GM CEO Mary Barra said in an NVIDIA press release announcing the deal.

DIG DEEPER: Unlock the value of generative artificial intelligence in your organization.

NVIDIA Announces GPU Advancements, Major and Partnerships

Huang made a number of other announcements during his address, including:

• Advancements in its graphics processing units over the next several years. NVIDIA’s latest Blackwell chips, announced earlier this year, are shipping now, he said. An upgrade to Blackwell, called Ultra, will become available in the second half of the year in the company’s upgrade. By the second half of 2026, NVIDIA will introduce its next generation of chips, which will be named for the late astronomer Vera Rubin. The Blackwell GPU is orders of magnitude more powerful than NVIDIA’s Hopper GPUs, and Rubin will be orders of magnitude more powerful than Blackwell. In fact, Huang characterized the Hopper GPUs as all but obsolete. “When Blackwells started shipping, you couldn’t give Hoppers away,” he said. “My sales guys are like, ‘Oh no, don’t say that!’ Look, there are still certain circumstances where Hopper is fine. That’s the best thing I can say about Hopper.”
• A partnership with Cisco on an “AI factory” architecture. The companies will work together to help enterprises build and secure data centers to develop and run AI workloads. “AI can unlock groundbreaking opportunities for the enterprise,” said Chuck Robbins, chair and CEO of Cisco, in a statement. “To achieve this, the integration of networking and security is essential. Cisco and NVIDIA's trusted, innovative solutions empower our customers to harness AI's full potential simply and securely.”
• A partnership with Oracle to help organizations speed creation of AI applications. The companies said they would integrate Oracle Cloud Infrastructure and the NVIDIA AI Enterprise software platform to make AI tools and services more easily available. “NVIDIA’s offerings, paired with OCI’s flexibility, scalability, performance and security, will speed AI adoption and help customers get more value from their data,” said Oracle CEO Safra Catz in a statement.

Keep this page bookmarked for articles from the event, and follow us on the social platform X @BizTechMagazine and the official conference feed, @NVIDIAGTC.