Nvidia GTC 2026: Everything Jensen Huang revealed for the future of AI

The future of Disney sounds a little dim

Olaf is happily regurgitating random sayings, while Huang believes this is the future of Disney.

It sounds like little robotic Disney characters could turn into a Terminator situation.

Welcome Olaf!

Huang introduces Disney's robotic character Olaf. Which Nvidia's Omniverse has helped craft in joint collaboration with DeepMind and Disney.

Big partnership with Uber on the horizon

Nvidia announces that various cars, including Hyundai and BYD, as well as the ride-hailing service Uber, will use Nvidia's robotaxi-ready platform.

Nemotron Coalition

AI leaders join Nvidia and its Nemotron coalition to help each region create its own sovereign AI.

These partners include Black Forest Labs, Cursor, LangChain, Mistral AI, Perplexity, Reflection, Sarvam, and Thinking Machines.

Open models for everything

Nvidia's AI models will be used in various facets of everyday life, from autonomous vehicles to robotics.

Huang wants Nvidia to empower everyone to become a part of the AI revolution.

Nvidia wants to create the best possible version of its model Nemotron so that companies, governments, and individuals can create their personalized models.

NemoClaw

OpenClaw is now enterprise-ready, says Huang, with NemoClaw.

Now that Nvidia has partnered with OpenClaw, the company has added various things to the agentic system.

"We are now at the frontier of every single domain of AI models," Huang says.

Openclaw exceeded what Linux did in 30 years in just a few weeks

Huang commends the open-source tool OpenClaw, despite security concerns.

OpenClaw builds an AI agent that is capable of completing various tasks. "You could say its an operating system."

"OpenClaw has open sourced the operating system of agentic computers," Huang said.

This software has made it possible to create personal agents, like companies have made it possible to have personal computers.

Data centers in space?

AI is 60% of Nvidia's business. So, why not colonize space with AI data centers?

Nvidia announces its Space-1 Vera Rubin module, which will help create AI data centers in space. This computer is going out to space and will start data centers out there, says Huang.

While the conditions are rough out there, Nvidia is working on it with their partners.

Nvidia DXS

Nvidia revealed its DXS platform, which is a digital twin that aids the operation of AI factories. The DXS AI factory platform is enabling the ecosystem of partners to build "the largest computer the world has ever seen."

From chip company to AI factory maker

Nvidia came from its chip-making roots and is now establishing AI factories by creating complex systems.

But this wasn't without collaboration. With collaboration comes communication. Therefore, Nvidia has created Omniverse, which allows collaborators to meet virtually to design gigafactories.

In two years time...

Huang says that in just two years' time, with this architecture, Nvidia will take token generation from 2 million to 700 million.

Rubin and Groq chips

The Groq chip has 500MB, 150 TB/s SRAM bandwidth, while the Rubin chip has 288GB.

These chips are "uniting processors of extreme performance."

Vera Rubin unlocks hundreds of billions in revenue

Huang suggests that the latest system can unlock $150 billion in revenue opportunity.

Rubin Ultra

Nvidia also presented the "Rubin Ultra," which is yet another system designed specifically for AI data centers, which will be powering the AI that we'll use in businesses, civil service, and for personal use.

Rubin Ultra can connect up to 144 GPUs.

Nvidia's new Vera Rubin platform

Nvidia reveals its latest platform, the Vera Rubin, designed for AI data centers, and is integrated completely into software. "It's designed for AI agent systems," says Huang.

"When we think Vera Rubin, we think the entire system." The system is designed for agentic AI. This system is going to "pound on memory" and the storage system "really hard."

"We created a brand new CPU that's designed for extremely high single-threaded performance."

Nvidia's token costs are world-class

"Our cost per token is the lowest in the world, you can't beat it," says Huang.

"Nvidia's token cost is world-class." This is due to extreme code design.

Huang says that Nvidia is the "token king."

Demand for computing has increased by 1 million times since ChatGPT

Huang thinks that AI usage and computing has grown incredibly since 2023.

This is not only due to the demand for RAM and the implementation of AI everywhere.

Computing demand will be higher than $1 trillion in growth, Huang believes.

"Everyone in the world should be using ChatGPT"

Huang talks about ChatGPT being the inception of the generative AI boom.

"ChatGPT started the generative AI era," said Huang.

"I use it (ChatGPT) every single morning."

AI and its impact on the world

Nvidia and Huang see their platforms as a necessary component to everyday life.

"We're talking about AI physics or AI biology for drug discovery, AI agents for customer service and support of diagnosis, and of course, physical AI, robotic systems."

Nvidia, Microsoft, and OpenAI

Huang attributes early deployment at Microsoft Azure as the installation that helped catalyze the company’s close partnership with OpenAI.

The Nvidia CEO said that theres an ongoing deep collaboration with Azure, including work on the AI Foundry, accelerating Azure cloud infrastructure, improving Bing search performance, and optimizing deployments across Azure regions, calling the relationship strategically important.

Nvidia powering big names like Puma and Salesforce

Nvidia is preparing partnerships with developer-customers like Puma and Salesforce that have integrated its technologies, and framed its cloud strategy as bringing those customers to service providers by integrating NVIDIA libraries to accelerate workloads and land customers’ applications on cloud platforms.

Next generation of graphics technology with DLSS 5

We're looking at the future of graphics technology with DLSS 5, helping transform games like Resident Evil Requiem.

DLSS 5 combines computer graphics with AI to create incredible experiences in gaming, making "computer graphics come to life," says Huang.

Combining 3D graphics with generative AI, which is controlled through structured data, resulting in content that's "beautiful and controllable."

"Structured data is the foundation of trustworthy AI," Huang says.