Two weeks ago, the Chinese American Semiconductor Professional Association (CASPA) held its 2023 Dinner Banquet, complete with a "fireside chat" with Jensen Huang, CEO of Nvidia. Jensen had plenty of interesting answers for the audience and personal stories, including career tips for aspiring young professionals in the so-called "AI era."
Narrowing down focus to Nvidia's main business (GPUs), one audience member asked if, given the expansive developments on many fronts, like reconfigurable, in-memory and photonic computing, GPU architectures will remain at the top for AI workloads in the coming years. (We've included an embedded video of this segment of the fireside chat below).
The audience member, a co-founder of CASPA and an AI architect, asked, "In the next few years, do you think GPU architecture will still dominate? Do we still have a chance (to beat Nvidia)?"
"You have a chance, but not much," Jensen responded, "It is, in fact, so little it is incalculable."
Jensen continued on a more serious note, "I'm just kidding. There are always good ideas— in fact, if there is a good idea, we'll change to it. You know, we're not stubborn. Our life goal is not to build CUDA GPUs."
Jensen drew comparisons to transportation technologies, which employ various types of transportation to achieve the same goal of moving items around the globe, and how a company's mission can lead them to fundamentally different answers, even if operating in the same field. "As you know, the G in GPU originally stood for graphics," Jensen said, "And today, we do much, much more than graphics. We changed the mission. I just never changed the name!"
As Jensen says, "Our life goal is to solve computer problems that normal computers cannot." While this mission statement can certainly be applied to innovations in real-time graphics rendering like RTX and DLSS, it's also quite clear that this applies to Artificial Intelligence and Nvidia's near-uncontested leadership in that area. There's no doubt in the industry that Nvidia seized the hardware opportunities presented by AI like no one else in the industry was willing or able to prior.
Before concluding, Jensen states, "We call it a GPU, but it does very different things. We're changing all the time. So if you come up with a good idea, let me know about it!"
