Abstract:
Nvidia CEO Jensen Huang said that customer ordering demand for artificial intelligence (AI) chips continues to exceed the company's supply capacity. In an interview on the "BG2" podcast released Thursday, Huang revealed that new orders are pouring in "almost every week, if not every day."
Huang's comments come as technology stocks are undergoing a sharp sell-off, raising questions about how buyers will fund the massive expansion of the artificial intelligence industry. Over the past six trading days, the market value of the U.S. semiconductor sector has evaporated by approximately US$1.64 trillion; investors are increasingly concerned about whether funds used for AI infrastructure construction come from debt financing.
Huang Renxun said: "We continue to increase supply and order volume continues to grow."
Large cloud service providers, emerging AI cloud operators, and equipment manufacturers have all placed large orders. Huang Renxun said that customers predict demand at least a year in advance, and the largest customers have even planned two years in advance.
However, the capacity Nvidia allocates to these customers is far below their actual needs.
Huang Renxun said that factors such as power shortages, unfinished data centers and manufacturing parts are also restricting the company's speed in expanding supply.
Wall Street questions: Who will pay?
At the same time, investors are also questioning whether artificial intelligence companies can generate enough revenue to support their huge borrowings.
CoreWeave is saddled with about $50 billion in debt, with yields on some of its bonds approaching 16%, a sign of growing investor concerns about its financial health.
However, the Nvidia-backed cloud provider reported at the end of June that its revenue backlog stood at $104 billion.
Nvidia owns about 10% of CoreWeave and has agreed to buy its unsold computing power through 2032 - an arrangement that highlights the financial exposure Nvidia has in the industry.
In an interview on Thursday, Jen-Hsun Huang said CoreWeave was "growing stronger every day" and said he didn't see the need for additional financial support.
In the BG2 podcast interview, Huang pushed back on concerns that OpenAI and Anthropic must generate enough revenue to justify investing trillions of dollars in AI infrastructure.
"There is a misunderstanding about the current situation," he said.
Huang Renxun estimates that laboratories at the forefront of artificial intelligence account for about 25% to 30% of Nvidia's computing business. The rest of the demand comes from traditional computing, startups, and companies developing a variety of specialized applications.
“They don’t need trillion-dollar revenues to be successful,” Huang said.
Comments