: Nvidia seeks to lead gold rush into the metaverse with new AI tools
Nvidia Corp. firmly staked its claim to the metaverse Tuesday by rolling out several products meant to not only give developers the software to chart new virtual worlds but the equipment needed to provide the foundation for those worlds.
The Santa Clara-based chip maker, which is becoming more and more involved with software and services, announced a major expansion of its Omniverse Enterprise platform at the company’s virtual GPU Technology Conference, or GTC, on Tuesday. Omniverse, which was beta launched in December, is a simulation and collaboration platform which “serves as the connective tissue for physically accurate 3D virtual worlds,” according to the company.
In prepared remarks, Nvidia
Chief Executive Jensen Huang said in his keynote address that with Omniverse, “we now have the technology to create new 3D worlds or model our physical world.”
Nvidia said it is launching Omniverse Avatar, which generates interactive AI avatars based on the company’s technologies in speech, computer vision, natural language understanding, recommendation engines, and simulation. Also, the company is launching Omniverse Replicator, an engine that generates the vast amounts of data needed to train the deep neural networks that constitute AI.
Omniverse Enterprise, which costs $9,000 a year, is also available to customers in a 30-day trial program, Nvidia said.
“A constant theme you’ll see — how Omniverse is used to simulate digital twins of warehouses, plants and factories, of physical and biological systems, the 5G edge, robots, self-driving cars, and even avatars,” Huang said.
On the subject of self-driving cars, Omniverse could help Nvidia with its smallest product segment that had been played up heavily in past years: Automotive. Nvidia reported auto chip sales of $152 million compared with total revenue of $6.51 billion in the second quarter.
Using Omniverse Replicator, Nvidia announced DRIVE Sim, which closes the gaps and increases the fidelity between virtual and actual worlds to help autonomous cars better perceive and navigate their environment and respond to rapid changes. DRIVE Sim can also be paired with Omniverse Avatar that would allow not only for an AI chauffeur but also a concierge in one’s car, the company said.
Nvidia also announced that it is partnering with data-center infrastructure company Equinix Inc.
to pre-install and integrate the company’s AI tech into its data centers worldwide. Equinix is the first program partner for Nvidia’s LaunchPad product, the company said. LaunchPad allows users to run large data workloads using AI technology for such applications as cancer research, Nvidia said.
Data-center sales of chipsets have become the fastest growing segment of the chip sector with Nvidia, Intel Corp.
and Advanced Micro Devices Inc.
all jockeying for market share as seen in recent earnings reports. In late October, AMD continued to show strength in data-center sales with the segment that includes those sales gaining 69% to $1.91 billion, and even Intel turned around a string of declines to report a 10% rise in data-center sales to $6.5 billion.
Analysts surveyed by FactSet expect Nvidia to post a 44% rise for $2.74 billion in data-center sales when it reports earnings on Nov. 17 after the close of markets. Nvidia specializes in making DPUs, or data processing units, which differ from a CPU or GPU in that it’s a data-center focused system on a chip designed for the high-speed transfer of data needed for such applications like AI, machine learning, security, and telecom.
Nvidia announced a product that uses the company’s Bluefield 3 DPU, the Quantum 2, which Huang called “the most advanced networking platform ever built.”
“Quantum 2 is the first networking platform to offer the performance of a supercomputer and the shareability of cloud computing,” Huang said.