Nvidia triples quarterly revenue, but sales in China are worrying
A start-up driving the artificial intelligence revolution may be in turmoil, but the semiconductor supplier driving its innovations only seems to be going from strength to strength.
On Tuesday, Nvidia continued its streak of blistering quarterly results, driven by stellar sales of AI applications from chips called graphics processing units, or GPUs. Microsoft is using thousands of those chips to run calculations for OpenAI, the generative AI startup whose CEO, Sam Altman, was fired on Friday and quickly hired by Microsoft to head a new cutting-edge research lab.
Nvidia said revenue for the third quarter ended in October tripled from a year earlier to $18.1 billion, while profit nearly quadrupled to $9.2 billion. The revenue figure was nearly $2 billion higher than the company forecast in August and much higher than analysts expected.
The company’s lofty forecasts since May have pushed its market value above $1 trillion and provide evidence that Nvidia is reaping the biggest financial rewards of any company chasing AI capabilities.
And the breakneck pace shows no signs of slowing as Nvidia ramps up production of a highly sought-after chip, the H100, that has been in short supply. Partly as a result, the company on Tuesday forecast revenue of about $20 billion for the current quarter, well above average analyst estimates of just under $18 billion.
“We have increased supply significantly each quarter this year to meet strong demand, and we expect to continue to do so next year,” Nvidia Chief Financial Officer Colette Kress said on a conference call with analysts.
The company’s growth has exploded despite fierce competition from other major chip makers, a number of start-ups, and the in-house creations of some of its major customers. In the latest development, Microsoft introduced its first two homegrown chips last week, including a processor aimed at AI jobs.
But Nvidia’s decades-long partnership with Microsoft seemed to be growing stronger nonetheless. The two companies announced a deal last week to offer Nvidia’s extensive collection of AI software through Microsoft’s cloud service running on Nvidia GPUs. A massive supercomputer that the companies jointly developed also finished third in a six-monthly ranking of the world’s fastest systems.
Nvidia CEO Jensen Huang praised the relationship between the companies at a Microsoft event and said there is much more innovation to come.
“Generative AI is the most important platform transition in the history of computing,” he says. “It’s bigger than the PC. It’s bigger than mobile. It will be bigger than the internet.”
The move by OpenAI’s board to fire Mr. Altman – followed by Microsoft’s move to hire him – introduces “a level of uncertainty that could temporarily create volatility in the generative AI market, and Nvidia is not immune,” Jacob Bourne, an analyst at Insider Intelligence, wrote in a research note.
But other analysts said the heavy spending by Microsoft and many other companies to buy Nvidia’s GPUs for AI tasks was likely to remain steady.
Mr. Huang said on the call with analysts that countries such as India, Sweden and France have taken steps to build specialized data centers for AI instead of relying on U.S. computing services.
“We see them being built everywhere, in virtually every country,” he said.
One headwind concerns China and the impact of new US export restrictions, which also affect Saudi Arabia and Vietnam. Several Nvidia GPUs, including models it developed for China, can no longer be shipped there without an export license.
Nvidia expects sales to affected countries to fall significantly in the fourth quarter, Ms. Kress said, although the shortfall will be more than offset by strong growth in other regions. The company is working on other chips that could be sold in China under the latest restrictions, she said.