Saturday, 26 April 2025
Saturday, 26 April 2025
On March 18, 2025, Nvidia CEO Jensen Huang took the stage at the GPU Technology Conference (GTC) in San Jose, California, delivering a keynote that didn’t just showcase the future—it declared it had already arrived.
The event, which the The New York Times dubbed as the “Super Bowl of AI,” drew over 25,000 attendees, including some of the biggest names in tech, all eager to witness how Nvidia plans to reshape the landscape of artificial intelligence.
The showstopper was Blue, a small, boxy robot that waddled onto the stage, beeping and nodding like a droid straight out of Star Wars.
“Today we’re announcing something really, really special,” Huang said, revealing that Blue is the product of a collaboration between Nvidia, Google DeepMind, and Disney Research, reported Education Next.
More than just a gimmick, Blue is powered by two Nvidia computers and runs on Newton, a new physics engine designed to train robots with a level of realism never seen before.
About 12,000 people packed into the National Hockey League arena in San Jose, hanging on Huang’s every word. NVIDIA announced a portfolio of technologies to supercharge humanoid robot development, including Nvidia Isaac GR00T N1.
“Every single year, more people come, because A.I. is able to solve more interesting problems for more industries,” Huang declared, making it clear that AI’s dominance is no longer a prediction—it’s a reality, according to The New York Times.
Huang introduced Nvidia’s new line of AI chips, Rubin, built for raw power. The supercomputer Kyber, showcased at the event, will run on 576 of these chips.
But Huang didn’t stop there. Nvidia announced it is open-sourcing Groot N1, a foundation model for humanoid robots. “I told you the progress of our robotics has been making enormous progress,” he said.
Groot N1 is designed to accelerate robot development worldwide, offering a customizable platform for complex reasoning and skills.
The keynote hit another peak with the unveiling of Nvidia Dynamo, which Huang called “the operating system of an AI factory.”
Dynamo is engineered to push Nvidia’s hardware—like the Blackwell Ultra and the upcoming Vera Rubin chips—to their limits, tackling massive computational tasks with staggering efficiency.
With 570 terabytes per second of memory bandwidth, Huang called it “the most extreme scale-up the world has ever done.”
From robots that could transform industries to AI infrastructure designed for unprecedented speed and scale, GTC 2025 wasn’t just about announcements—it was a declaration of AI dominance.
