Defined by the powerful keynote highlighting the breakthrough partnership between Amazon Web Services Inc. and Nvidia Corp., there’s been a surge in the demand for powerful generative artificial intelligence powered by cutting-edge hardware and networking.

AWS, with a rich history of pioneering cloud-based Nvidia integration, stands as a leader in volume, setting the stage for an exciting future with the L40Ss graphical processing units and the application in digital twins.

“I’m really excited about the new L40Ss [GPUs] because that’s been a big area in mind where I work — the digital twin and a lot of the simulation, training and autonomous vehicle components, and what they’re doing with Omniverse is going to be the future for that in a lot of ways,” said Bill Vass (pictured), vice president of engineering at Amazon Web Services Inc. “We’re integrating it with TwinMaker we announced today. So you’ll that coming as well.”

Vass spoke with theCUBE industry analysts Dave Vellante and John Furrier at the “Supercloud 5: The Battle for AI Supremacy” event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the burgeoning AI landscape as it will be shaped by strategic partnerships, technological innovation and a commitment to sustainability.

Digital Twins and the Omniverse revolution

A digital twin denotes a digital representation of a physical object, person or process in the context of a digital version of its environment. Digital twins can help companies simulate real situations and any respective outcomes, for instance.

“We’ve got thousands of customers building digital twins and the future, what we’ve been building up for over the years is sort of this virtual cycle, and we always like flywheels at Amazon — everything flywheel around here,” Vass said. “You got S3, you got the new accelerated S3, it’s very exciting. You’ve got FSx, you’ve got all these ways to manage exabytes and in the future zettabytes of data.”

The integration of Nvidia’s Omniverse, boasting 100% ray tracing and full physics capabilities, marks a significant leap forward. The real-world applications, ranging from Amazon’s fulfillment centers to the announcement of the TwinMaker, propel digital twins into a new era.

“Everything we’ve been doing the last 10 years or so has been building up all of the parts,” Vass said. “And I wish we could’ve done it right out the gate, but the technology wasn’t due there. Elastic Fabric Adapter wasn’t there to be fast enough [as] interconnects for memory sharing at the time — we didn’t have Graviton, we didn’t have Trainium and we didn’t have the level of GPUs we have today — So all of that is coming together.”

The generation of synthetic data is an emerging frontier crucial for training models without real-world constraints. This, combined with the shift toward software-defined infrastructure, laid the groundwork for the digital twin revolution.

“Once you have all that, you can apply machine learning and these new LLMs are going to be able to do things like generate a lot of that synthetic data,” Vass said. “They’re not quite there yet. And in one area, we really need them to generate 3D data and I see a new paper every week getting closer and closer to that.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the “Supercloud 5: The Battle for AI Supremacy” event:

Photo: SiliconANGLE