If AI fails to drive revenue or efficiency, this cycle of reinvestment could

Speaking on CNBC Friday, Huang stated that the AI buildout has "seven to eight years to go" and emphasized that "AI is going to fundamentally change how we compute everything.

The tech executive noted that demand for Nvidia’s products is "sky high" and pointed out that GPUs sold six years ago are increasing in price, highlighting the sustained market interest in AI hardware.

Huang specifically mentioned that AI companies like Anthropic and OpenAI are "making money" but remain "computer constrained," indicating they need more computing resources to expand their operations. He added that "there’s no drama with OpenAI" and stressed that the company needs Nvidia’s new generation chips. 

Regarding competition, Huang acknowledged that China is undoubtedly a competitor but maintained that it "makes no sense to concede China market if you want to win globally."

When discussing companies utilizing AI technology, Huang singled out Meta, saying "no one uses AI better than Meta." He credited their ability to deploy AI at a massive scale to improve user engagement and advertising efficiency.

 He also mentioned that his biggest concern is "AI being effective." because the current $660 billion infrastructure buildout relies entirely on AI delivering tangible economic utility.

His concern centers on several key factors:

• Sustaining the "Virtuous Cycle": Huang believes that as long as AI companies generate profit and real-world results, they will continue to "double down" on investments. If AI fails to be effective (i.e., fails to drive revenue or efficiency), this cycle of reinvestment could stall.
• The Shift from Hype to Utility: He noted that the market is transitioning from "AI hype" to "AI utility". The massive capital expenditures by "Hyperscalers" like Meta, Amazon, and Microsoft are only sustainable if these tools fundamentally improve business outcomes, such as Meta's success in using AI for advertising efficiency.
• Real-World Application Layer: Huang has long argued that the "top layer" of the AI stack—the applications in healthcare, manufacturing, and financial services—is where the actual economic benefit happens. His concern is ensuring these models are "good enough" to transform these traditional industries effectively.
• Productivity vs. Job Loss: He warned that if AI is not used to generate "fresh ideas" and new value, productivity gains could simply lead to job losses rather than economic expansion. Effectiveness, in his view, means AI empowering workers to tackle more ambitious projects rather than just replacing them