The tech industry’s surge in capital spending on AI infrastructure is justified, appropriate, and sustainable. Nvidia CEO Jensen Huang said Friday on CNBC’s “Halftime Report.”
“The reason is that the cash flow of all these companies will start to increase,” Huang said.
Nvidia stock rose 7% during Friday trading.
Huang’s comments are aimed at NVIDIA’s major customers meta, Amazon, google and microsoft Reported the latest earnings for the past two weeks. These companies have told investors that they plan to significantly increase spending on AI infrastructure. These hyperscalers could spend a total of $660 billion in capital expenditures this year, with much of that spending going toward buying Nvidia chips.
Wall Street’s reaction to the spending spree was mixed, with shares of Meta and Alphabet rising, while Amazon and Microsoft were punished.
Huang said the “largest infrastructure build-out in human history” is being driven by “very high” demand for computing power that AI companies and hyperscalers can use to make more money. He gave specific examples of what Nvidia customers are doing with AI.
Meta is using AI to move from a recommendation system that ran on CPUs to one that uses generative AI and agents, Huang said. He said Amazon Web Services’ use of Nvidia chips and AI will impact how the retail giant recommends products, and that Microsoft will use Nvidia-powered AI to improve its enterprise software.
He also praised OpenAI and Anthropic, two major artificial intelligence labs that are using Nvidia chips through their cloud providers. Nvidia invested $10 billion in Anthropic last year, and Huang said earlier this week that the company would make a significant investment in OpenAI’s next round of funding.
“Anthropic is making big profits. Open AI is making big profits,” Huang said. “If you have twice the computing power, you’ll increase your revenue by a factor of four.”
Reflecting continued demand for AI computing power, all graphics processing units that Nvidia sold in the past, including six-year-old chips such as the A100, are now being rented, he said.
“As long as people keep paying for AI and AI companies can make money from it, companies will keep doubling, doubling, doubling, doubling,” Huang said.
