Nvidia CEO breaks down the ‘layers’ of AI & how they work together

Yahoo Finance Video

December 4, 2025

In this video:

•                                       StockStory Top Pick 
  

  NVDA

  +2.50%

Nvidia (NVDA) is probably the biggest name in AI. But AI takes energy, large-language models, and lots of data to work. Speaking at a Center for Strategic and International Studies event, Nvidia CEO Jensen Huang explains four layers of AI and how they work together to help create the technology.

Watch more expert insights and analysis on the latest market action.

Video Transcript

00:00 Speaker A

Now, the way that that AI works and and uh our technology works is that in the final analysis, the technology platform is built in layers and that’s one of the reasons why we think of it as a platform. You’re standing on top of it. An application or an industry stands on top of that platform. That platform starts with energy on the bottom.

00:23 Speaker A

One of the reasons why why uh this administration uh has made such a huge difference right away if it’s is a pro energy growth initiative. Its attitude about energy is that if we don’t have energy, we can’t enable this new industry to thrive. It is absolutely true.

00:43 Speaker A

So layer one is energy. Layer two are essentially the chips and systems, but the chips. That’s where Nvidia comes in. Layer three is a whole bunch of software and we build a whole bunch of software on top of our chips. And we’re well known for this piece of software called Cuda, but there’s hundreds of different pieces of software that we create that enables people to do AI for different fields of science or language or images or whatever happens to be robotics and manufacturing and such.

01:16 Speaker A

But that third layer is called infrastructure, basically software. Now, people historically have thought of infrastructure really as cloud. But increasingly, it’s really important to realize that infrastructure includes land, power, shell, because these are and I’ll speak about this in a second, that this industry spawned another industry altogether and I’ll come back to that.

01:40 Speaker A

But the third layer is basically infrastructure and that infrastructure includes financial services because it takes an enormous amount of capital to do what we do. and historically all of that software. The layer above that, and this is where people largely focus on when they talk about AI, which is the AI models.

02:02 Speaker A

This is the revolutionary, of course, chat GPT, uh incredible work that Anthropic does with Claude and uh Google does with Gemini and uh what XAI does with Grock.

02:16 Speaker A

But the important thing to realize, those are four of the one and a half million AI models in the world. Mm.

02:27 Speaker A

AI is not just intelligence that understands English or language, but it’s AI that understands genes, proteins, chemicals, the laws of physics, AI that understands quantum, AI that understands physical articulation, otherwise known as robotics, AI that understands patterns across a long sequence of time, financial services, AI that understands longitudinally across multiple modalities, healthcare.

03:05 Speaker A

And so AI has all of these different reaches and domains. We talk about this one area, we just have to be very careful to understand that AI spans basically every form of information across every field of science, across every single industry.

Terms and Privacy Policy

Privacy Dashboard

More Info