At its annual developer conference in San Jose, California, Nvidia CEO Jensen Huang unveiled groundbreaking advances in artificial intelligence technology. The spotlight was on the upcoming Blackwell Ultra GPU, designed to support larger AI models with expanded memory. Set for release in the second half of 2025, this chip aims to deliver enhanced performance for both training and real-world inference, despite current manufacturing delays due to a design flaw.
In another innovative step, Nvidia introduced the Vera Rubin computing system. This cutting-edge platform combines a custom-designed processor with next-generation GPUs to tackle hyperscale AI workloads. Named in honor of pioneering astronomer Vera Rubin, the system promises superior chip-to-chip data transfer speeds and is scheduled for launch in late 2026, with follow-up models planned for 2027 and 2028.
These advancements underscore a dynamic shift in the AI landscape, as industry players pivot from simply training systems toward deploying them effectively in the real world. Nvidia's commitment to pushing technological boundaries continues to resonate with tech enthusiasts, business innovators, and global changemakers looking for fresh, impactful insights into the future of artificial intelligence.
Reference(s):
Nvidia unveils updated AI solutions amid design flaws, industry shifts
cgtn.com
