COMPUTEX 2023 was off to a roaring start on Monday, with an Opening Keynote delivered by Jensen Huang, founder and CEO of the world-leading American tech company NVIDIA, that pressed home how rapid advancements in accelerated computing and generative AI are driving a revolution that promises to impact every corner of the industry and beyond.
Speaking for two hours to a packed hall at the international trade show organized by the Taiwan External Trade Development Council (TAITRA), the key theme of Huang’s speech was that dual innovations in accelerated computing and generative AI are enabling NVIDIA’s supercomputers to “write and develop software that is incredible, and that can be deployed in devices all over the world.”
The age of traditional CPU-based computing, in which computing power could be expected to increase tenfold every five years, is now ending.
“Accelerated computing is the path forward,” said Huang in a speech that detailed how the technology has now reached a tipping point and is already being utilized for every kind of application imaginable.
Likewise, AI is characterized by its ability to process and transform information across many different modalities, vastly expanding its range of utilizations and applications.
Huang illustrated the technology’s capabilities with a demonstration of NVIDIA’s new Avatar Cloud Engine, which integrates speech recognition, natural language competence, and facial animations and gestures in response to voice cues to render and animate an unscripted avatar in real time.
This paradigm shift is changing the way we think about the cost of high-performance computing, Huang explained as he introduced NVIDIA’s HGX H100 GPU server platform, “the world’s single most expensive computer for which you can say: the more you buy, the more you save.”
Today’s computing networks are structured on a framework where the data center itself is the computer, which is why NVIDIA are working to extend their advancements in accelerated computing and generative AI to the cloud.
In their vision, “every data center can be an AI data center,” said Huang, which would bring the technology to virtually limitless applications, so that “every industry can now benefit from generative AI.”
He demonstrated with the example of how manufacturing and heavy industries could utilize NVIDIA’s new Omniverse Cloud service, running on Azure from the company’s servers in California, to visualize, design, and simulate a factory floor and production line, with multiple users collaborating from around the world in real time.
In another example, the company announced a partnership with WPP, the world’s largest advertising services agency, to build content generation engines based on the Omniverse platform, which could be used to generate precise advertisements for individual users.
“In the future, when you engage information, much of it will be generated,” Huang said. “The computing model has changed.”