By Marty Swant  •  March 20, 2025  •

“The way you access data will be fundamentally different from the past,” he said, citing Perplexity’s model of fusing traditional search methods with conversational AI as the new norm. “Instead of doing retrieval that way, I just ask Perplexity what I want… This is the way enterprise IT will work in the future as well.”

Among the launch was an open-source software called Dynamo, which is designed to improve the efficiency and scalability of large language models. Dynamo optimizes GPU resource allocation to accelerate AI processing, reduce costs, increase speed and enhance real-time inference, according to Huang. He added that Dynamo will process computational resources to produce AI tokens — the building blocks of AI content and enterprise solutions.

Although Huang didn’t mention too many examples on the Nvidia GTC 2025 stage, partners mentioned in a blog post about Dynamo include AWS, Google Cloud, Microsoft Azure, Meta, and Perplexity AI.

Monks showcases Nvidia tech

Nvidia isn’t a typical advertising-focused tech giant, but major brands and marketing agencies increasingly partner with it to power their own capabilities. One example is Monks, which announced a new 50-person team called the Monks Agentic Advisory that operates as a “nimble advancement and consulting team.” Monks is also certifying 150 engineers to build and deploy custom generative AI models for clients.

To showcase how it uses Nvidia tech, Monks also unveiled a new AI-generated video for Puma that uses agentic AI to create hyper-realistic content using Monks workflows and generative video tools from AI-startup Runway.

Monks Chief Innovation Officer Henry Cowling said agencies and clients have to consider how AI will impact how people work, workflows, and broader conversations about how organizations are designed. “I don’t think you can talk seriously about enterprise transformation or specifically marketing operations transformation without that conversation leading to Nvidia,” Cowling said. “At some point, but even if it’s just at the level of the chip stack, eventually you hit Nvidia and they’re obviously working migrating their way sort of up value stack as well.”

More Nvidia GTC 2025 announcements

• T-Mobile: T-Mobile and Cisco will work to develop AI-native wireless capabilities designed for 6G wireless, with Huang claiming AI could revolutionize communication by mimicking how humans use context and prior knowledge to convey meaning.

• Oracle: Another partnership will see it work with Oracle to help companies more easily organize and search through large volumes of text, images and videos.

• Google: Nvidia will be the first to adopt SynthID, a Google DeepMind AI technology that embeds watermarks into AI-generated content to enhance content transparency and trust in generative AI by safeguarding against misinformation and misattribution.

• Accenture: The consulting firm announced a new AI agent builder that will use the newly released Nvidia Llama Nemoton reasoning models. In a statement about the news, Accenture also said it’s already built 50 industry-specific AI agents for clients like ESPN, HPE and the United Nations.

• Disney: Nvidia and Google Deepmind previewed a collaboration with Disney to create Star Wars-like droids. To develop the robots, the companies are using numerous new Nvidia innovations including Newton, an open-source new physics engine that simulates robotic movement in real-world settings.

https://digiday.com/?p=572376