It's hard to find a company that has benefited more from the AI boom than Nvidia (NVDA). The company saw revenue grow 262% in the first quarter. Its data center division in particular grew 427% year over year to $22.6 billion. The reason? Nvidia founder and CEO Jensen Huang says the company is essentially creating “AI factories.” “These AI factories are basically what we're building. We build it as a comprehensive unit, a comprehensive architecture and platform, but then we break it down so that our partners can deploy it in any kind of data center,” Huang told Yahoo Finance in an exclusive interview.
Regarding demand for these products, Huang claimed it was “very strong,” adding, “After we announced Blackwell, demand for Hopper grew throughout the quarter, which shows how much demand there is. People want to deploy these data centers right now.”
Be sure to check out our full interview with Nvidia CEO Jensen Huang.
This post was written by Stephanie Mikulich.
More on Nvidia from Yahoo Finance:
Nvidia shares rose 4% after the company reported better-than-expected profits, a stock split and a dividend hike.
Nvidia CEO says Tesla is 'far ahead' in self-driving tech as self-driving efforts boost chip demand
Why this analyst says Nvidia isn't a stock to buy
Nvidia is 'at the center of the growing tech story': Strategist
Beyond the Ticker: Nvidia
Video Transcript
You mentioned on several occasions on the phone that supplies of both Hopper and Blackwell chips would be limited.
Well, until next year, because demand is so high.
Well, what can you do about that?
Demand has increased throughout the quarter following the Blackwell announcement, but are there levers you can leverage to increase supply?
That means we know how much demand there is.
People want to deploy these data centers right now.
They want to get our GP US up and running right now and start making money and saving money.
So the demand is very strong.
Well, it's really important to take a step back and realize that we're not making GP U chips, we're making what we call Blackwell and GP U.
But we are actually building an AI factory.
These AI factories have CP US and GP US and have very complex memories.
The system is really complex.
This is connected by MV Links, we have MV Link switches, we have Financial switches, we have Ethernet switches and Ethernet N, and all of this is connected by a very complex spine called MV Links.
And the amount of software required to build and run all of this is incredible.
So these AI factories are essentially what we build, we build them as a holistic unit, as a holistic architecture and platform.
But then we break it down so that our partners can take it and put it in any type of data center.
And each cloud has a slightly different architecture and stack.
And so our stags and our architecture can become deeply integrated with theirs.
But everyone is a little different.
So we're going to build this as an AI factory and then we're going to break it down so that everyone can have an AI factory.
This is really great, and we do this very diligently and in very large quantities.
But it’s extremely difficult to put it into practice.
So every component, every piece of our data center is the most complex computer the world has ever made.
So it makes sense that almost everything is restricted.
