Recently, Nvidia CEO Jensen Huang said during a keynote that Nvidia’s GPUs are “so good that even when competitors’ chips are free, it’s not cheap enough.” Huang also explained that Nvidia GPU prices are not that important in terms of the total cost of ownership (TCO) of an AI data center.
The impressive scale of Nvidia’s achievements in powering the booming AI industry can’t be denied—the company recently became the world’s third-largest market capitalization thanks in large part to its AI GPUs—but Jensen’s comments are sure to be controversial as he disparages all of the solutions of competitors like AMD, Intel, and a number of competitors with ASICs and other types of specialized AI silicon.
Starting at 22:32 in the video above, John Chauvin, former director of Trione SIEPR and professor emeritus of economics at Stanford University asks:
You create cutting-edge chips. Is it possible that you’ll run into competition that claims to be good enough – not as good as Nvidia, but good enough and much cheaper? Is this a threat?
Jensen Huang began his response dramatically:
We have more competition than anyone on the planet.
He told Chauvin that even Nvidia’s customers are its competitors in some cases. In addition, Huang noted that Nvidia is actively helping customers developing alternative AI processors, going so far as to reveal to them which upcoming Nvidia chips are in development.
This does sound like a rather unusual way of doing business, but Huang’s next assertion, that Nvidia operates as a “completely open book”, partnering with virtually everyone else in the industry, may be harder to believe. More recently, there have been serious allegations that Nvidia is running a “GPU cartel” and it has been reported that customers have been afraid to talk to rival GPU/AI accelerator makers due to fears of backlogged orders. An industry consortium has also been created in an attempt to shift the company’s dominant CUDA programming model.
Returning to the interview, Jensen Huang outlined Nvidia’s unique advantages that are unshakable today. Nvidia’s CEO said that while it’s possible to create a chip that works well with one specific algorithm, Nvidia’s GPUs are programmable. What’s more, Nvidia’s platform is “a great standard…in every cloud computing company.” So the typical data center that wants to support a wide range of clients, from financial services to manufacturing and beyond, will be drawn to Nvidia hardware.
Huang also tried to make a comparison that might seem counterintuitive to those focused on GPU prices: People who buy and sell chips think about the price of the chips, while people who run data centers think about the cost of operations, he explained.
Of course, companies will be keenly aware of total cost of ownership (TCO), which basically means that Nvidia’s stated benefits of deployment time, performance, recycling and flexibility are “so good that even when competitors’ chips are free, it’s not cheap enough.” , according to Huang.
Concluding his response to Chauvin, the Nvidia CEO emphasized that staying far ahead is Nvidia’s goal. Huang reminded summit attendees that what Nvidia does requires a lot of hard work and innovation—there’s no luck involved, and nothing is taken for granted.
Naturally, Nvidia’s competitors would not agree with Huang’s statements, but so far Nvida really has no real significant competitors in the field of AI.
2024-03-11 16:40:00
#Jensen #Huang #free #chips #competitors #beat #Nvidia #GPUs
