Follow Datanami:
August 9, 2022

Nvidia Could Become Google of the Metaverse with AI Search Engine


Nvidia is showing signs that it wants to be the Google of the next-generation 3D Internet, which is being projected as an upgrade from the current 2D Internet.

The company has developed a contextual search engine for three-dimensional (3D) data called DeepSearch, which it could bring to the public Internet in the future. The search engine is targeted at 3D data, which could be images, graphical assets, and other contextual data.

DeepSearch is an “AI powered, modern and scalable way of searching the vast amounts of 3D data that was previously dark for many enterprises and people around the world in their catalogs,” said Rev Lebaredian, vice president of Omniverse and simulation technology at Nvidia, in a call with media.

Nvidia is emerging as a major player to create the 3D version of the Internet, which is widely referred to as the metaverse. The metaverse is seen as a virtual universe where avatars can work, play, and interact. Games like Minecraft and Fortnite, and virtual reality platforms like Roblox, are seen as early instantiations of the metaverse.

“We’ve built the core technology that allows us to search 3D data which was previously unsearchable. You can use natural language text to search. You can use images to search for things that are similar inside the 3D catalog,” Lebaredian told Datanami.

Nvidia is enabling AI-powered visual and text search with its DeepSearch engine

Facebook is betting its future on the metaverse as its once-hot 2D-based social media assets lose ground to platforms like Tiktok. Microsoft is buying gaming company Activision Blizzard for $68.7 billion to grow in the metaverse business.

Nvidia is sharing more information about DeepSearch at the SIGGRAPH conference this week in Vancouver, Canada, where it announced availability of the software to its customers.

“Over time, we’ll be making this technology available publicly as well through various other services. That’s to be decided. What we’re showing and announcing here is the core technology. How it ends up reaching everyone, that’s going to evolve over time,” Lebaredian said.

DeepSearch will initially be available for its Omniverse customers to install on premises within enterprises without exposing it to the public at large, Lebaredian said. Omniverse is an overarching brand used by the company to group its 3D software offerings.

Nvidia last year described DeepSearch as a search engine for a 3D asset library that can be searched via text or by dragging or dropping an image. That feature makes it easier for game developers to search for 3D graphics that it can drag and drop in environments.

The DeepSearch spider uses AI technology to scour through tags of 3D models, and automatically adds contextual tags by searching through the entire content associated with 3D models. The system uses deep learning to classify, index and retrieve objects in 3D models.

The idea of a search engine for the metaverse isn’t new. Ludo focuses on Web3 concepts such as NFTs and cryptocurrency, which are considered foundations for e-commerce in the metaverse. Canadian firm Lighthouse Labs is creating a metaverse search engine to “discover places, friends, experiences and events across the spatial web,” the company said on its website.

Nvidia’s search engine enriches discovery by digging deeper to extract context out of assets in the 3D world.

Nvidia is using SIGGRAPH as a launchpad for many metaverse offerings. The metaverse is changing the way Nvidia thinks about graphics creation and distribution, and is inserting a healthy dose of AI to make graphics delivery faster and more efficient.

Related Items:

Nvidia CEO Touts a ‘Million X’ Speedup in AI

Nvidia Launches Hopper H100 GPU, New DGXs and Grace Megachips

Thinking in 10x and Other Google Directives