Nvidia CEO Jensen Huang and the future of…everything

Nvidia CEO Jensen Huang confirmed his reputation as a futurist, predicting every single car, truck, cell tower—indeed every single edge device—will in effect be a data center in a decade.

He also said in a call with reporters on Tuesday that the chip shortage hurting automakers will sort itself out in a couple of years.

The concept of edge devices acting with the capabilities of a data center might not be completely new, but Huang cemented it.

“Every single data center will have its infrastructure computing platform isolated from the application platform in five or 10 years,” he told reporters as part of the company’s GTC21 event.  “It’s going to be complete. Every single edge device will be a data center…Every single cell tower will be a data center, every base station…Every single car… truck, shuttle will be a data center.”

His comments came a day after he delivered a 90-minute keynote at the company’s GTC21 filled with announcements of a slate of new technologies, many designed to serve future needs of AI. Wearing his signature leather jacket and speaking from his kitchen, he described a processor named Grace, named after the pioneering Grace Hopper, along with an AV SoC called Atlan and other hardware and software tools. Some 180,000 engineers and designers registered for the digital event that included 1,600 tech talks.

RELATED: Nvidia GTC21 launches could further ‘democratize’ AI for enterprises

In future data-center-like vehicles the application plane will be isolated from the control plane, which will function in the chassis of the vehicle, he explained. That isolation will mean the control plane will be “functionally safe” from attacks, he said, enabled with the help of Nvidia’s new Bluefield product.

The significance of Huang’s comments is that security will continue to be (if not already is) paramount for all devices. Imagine having every single edge device, such as robots in a factory, and not have the factory completely tamper proof, he postulated. “Everywhere on the edge autonomous machines will be isolated,” he said.

Huang also predicted Grace, the company’s first data center CPU announced on Monday, will represent a “very large new market” because of the immense need to crunch data with AI in coming years for things such as natural language understanding, and the desire to create a digital twin of the entire Earth for accurate weather forecasts and other needs. Nvidia refers to digital twins created in a virtual world as part of a metaverse.

RELATED: BMW features digital factory twin at Nvidia GTC21

Asked what he makes of the current chip shortage, which has been especially difficult for carmakers, Huang was not alarmed.  

“I doubt [the chip shortage] will be a real issue in two years,” he said. “The industry recognizes this and will build out the necessary capacity.”

RELATED: Chip shortage: How soon is relief possible?

He said the chip industry has been caught in a couple of dynamics, including the pandemic and the weakness it exposed in the supply chain for autos.

“There are too many components in cars, with various supply chains and it is super complicated,” Huang said. Cars have to be re-architected, he said, with a few centralized components, perhaps four components instead of thousands.

After the abrupt shutdown because of Covid-19, the “restart was far more complicated than anybody imagined,” he said.

When carmakers stopped ordering chips because auto demand was low a year ago, the chip industry started focusing on advanced computing needs. Many analysts agree that when carmakers started needing chips last fall with sudden orders of more cars from consumers, chip manufacturers were busy making other chips and could not convert quickly to making auto chips.

At GTC21, Nvidia also announced Atlan, a system-on-chip that will offer 1000 TOPS of performance for autonomous vehicle use in models introduced in 2025. 

Huang has “every confidence” that Atlan chips will be available for carmakers, who are already designing their 2025 models and will be able to sample Atlan in 2023. “Nvidia …works with all the major chip foundries,” Huang said. “We’ll do whatever is necessary to do when the time comes. A company of our scale and resources can surely adapt the supply chain and make [Atlan] available to customers that use it.”

RELATED: Nvidia shows Atlan processor for AVs along with sim and security tools