Keynote by NVIDIA CEO Jensen Huang | NVIDIA GTC 2025
We watched NVIDIA CEO Jensen Huang deliver a truly ground-breaking GTC keynote. Here's what you need to know.
Welcome to GTC 2025.
This year Nvidia GTC keytone was truly inspiring and ground-breaking.
Watching Jensen Huang's latest announcements, I'm struck by how NVIDIA is redefining the entire AI landscape—not just incrementally improving technology, but reinventing how we think about computation itself. Here's my synthesis of their groundbreaking innovations.
Overview of topics covered:
GeForce RTX 5090
"It all started with GeForce."
GeForce is sold out all over the world.
The GeForce RTX 5090, NVIDIA's most powerful consumer GPU to date, represents what Jensen Huang calls "the most significant computer graphics innovation since we introduced programmable shading 25 years ago."
According to Huang, the RTX 5090 delivers twice the performance of the previous generation RTX 4090, powered by the combination of Blackwell architecture innovations and DLSS 4
AI Real-Time CGI Rendering
Role of artificial intelligence in computer graphics is on the all time high. AI predicts additional pixels for real-time rendering. Temporal stability is very important in graphics rendering.
Jensen Huang highlighted AI Real-Time CGI Rendering as a revolutionary advancement in computer graphics. He described it as "100% path traced for every pixel that's rendered" where "artificial intelligence predicts the other 15".
Partnership with GM
Announced the partnership with GM for self-driving car fleet.
Focus on AI for manufacturing and enterprise. AI infrastructure development for GM's vehicles. Emphasis on automotive safety technology.”
"It's called Halos." Safety requires technology from silicon to system software.
"Nvidia is the first company in the world... to have every line of code safety assessed."
Blackwell GPU Data Centers
Transition in computer architecture and scaling up.
He predicted that we'll soon see massive AI factories powered by Blackwell GPUs that will be "visible from space," with data centers housing up to 1 million GPUs. These million-GPU clusters would interconnect with other similarly large data centers in proximity, creating extensive data center complexes.
Nvidia Dynamo
Announcement of Nvidia Dynamo as the operating system of an AI factory.
Transition from traditional data center operating systems like VMware.
Future focus on agents rather than enterprise applications.
Nvidia Robotics
"The time has come for robots."
Robots can interact with the physical world. There is a severe shortage of human workers expected by the end of the decade.
Nvidia Omniverse
"We created a system called Omniverse." It is the operating system for physical AI. New technologies added to scale AI with generative capabilities. Cosmos generates an infinite number of environments. It allows for controlled and grounded data creation.
Showcased control over robots in various scenarios. Omniverse is essential for creating those environments.
Partnership between Deep Mind, Disney Research, and Nvidia called Newton.
Focus on training robots with super real-time simulation. Groot and Future of Robotics
"Groot is open sourced."
Future training of robots will utilize advanced simulations.
Thanks for your time!
More detailed summary to follow soon..