By Sanjit Singh Dang, PhD
Publication Date: 2026-04-21 03:56:00
Nvidia CEO Jensen Huang introduces Vera Rubin, a next-generation AI data center platform during the keynote address at the company’s annual GTC developers conference in San Jose, California, on March 16, 2026. (Photo by JOSH EDELSON / AFP via Getty Images)
AFP via Getty Images
At Nvidia’s 2026 GTC conference in San Jose, Jensen Huang did something unusual even by Silicon Valley standards. He did not just outline a roadmap. He quantified the future of artificial intelligence. He declared, “I believe that computing demand has increased by one million times in the last two years.”
Huang now projects at least one trillion dollars in demand for Nvidia’s Blackwell and Vera Rubin systems through 2027, doubling the company’s previous $500 billion estimate from just a year ago. But in the weeks since that announcement, new details have made one thing clear. The headline number is already becoming outdated. This is not a peak forecast. It is a moving target.
AI Acceleration is the Real Story
The most important update is not that Nvidia sees one trillion dollars in demand. It is how fast that number is changing. At GTC, Huang emphasized that compute demand has effectively gone “off the charts,” describing growth that has increased by orders of magnitude in just a few years. What this means is that while the one trillion dollar demand number looks huge, it may get upgraded again in a few months.
That acceleration is now visible across the entire stack. Nvidia is no longer…
