U.S. Tech Stocks Drop As AI Uncertainty Knocks Chip Groups
By: George Steer and Aiden Reiter in New York and Michael Acton in San Francisco
Financial Times
US tech stocks slipped on Friday as weak earnings at a large chip designer fed investors’ concerns about slowing growth across the artificial intelligence sector.
The tech-heavy Nasdaq Composite finished the day down 1.2%, with semiconductor group Marvell Technology falling 18.6% after its data center revenue missed analysts’ forecasts. Bank of America on Friday downgraded the stock to “neutral” from “buy”.
Nvidia, the $4T semiconductor giant that this week signaled uncertainty around its sales in China fell 3.3%.
Friday’s declines dragged the Philadelphia Semiconductor index, which tracks 30 of the world’s biggest semiconductor manufacturers, to its worst performance since mid-April.
Wall Street’s S&P 500 fell 0.6%, its biggest one-day drop since the start of August. Analysts attributed some of the declines to investors taking some profits on the last trading day of the month, with the benchmark index ending August up 1.21%.
Tech stocks have powered US equity markets to record highs, yet simmering concerns about surging spending on AI infrastructure have in recent weeks weighed on some of Silicon Valley’s biggest names.
The slide comes as Nvidia, the world’s biggest listed company, on Wednesday gave a revenue forecast that failed to meet investors’ highest expectations, and narrowly missed analysts’ estimates for its data center revenue last quarter.
Nvidia’s shares fell after it also cited uncertainty around US export controls, which have hampered its AI chips business in China.
“Though [Nvidia] didn’t exactly disappoint expectations, concerns about its China penetration and forward guidance somewhat dampened enthusiasm,” said Florian Ielpo, head of macro at Lombard Odier Investment Managers.
Hundreds of billions of dollars have already been spent on the data centers required to power generative AI models such as ChatGPT, which investors hope will usher in a new era of rapid productivity growth.
But generative AI revenues at the so-called hyperscalers — Amazon Web Services, Microsoft Azure and Google Cloud — were just $45B last year, according to Morgan Stanley.
Marvell, which competes with Nvidia by building custom chips for the likes of Amazon and Microsoft, saw its shares surge last year on the wave of investor enthusiasm around AI hardware.
But they have fallen more than 40% since the start of the year, as it has been hit by trade uncertainties, leading it to cancel an investor day earlier this year, as well as questions about whether it can grow its client base.
Marvell’s warning of slowing growth in its data center business came as shares in Chinese ecommerce giant Alibaba jumped on reports that it is set to release a new semiconductor chip, as part of a Beijing-backed effort to displace Nvidia’s popular H20 processors.
Nvidia reached a deal with the Trump administration to resume exports of its H20 chip to China earlier this year in exchange for the US government receiving 15% of the revenue from the sales. It said on Wednesday that it was waiting on the administration to codify the plan.
The Chinese government has discouraged domestic AI companies from buying the chip, however, and is increasing its efforts to compete with the US chipmaker.
Earlier this week Chinese AI chipmaker Cambricon posted record profits for the first half of the year while announcing several technological improvements that it says bring its chips closer to the quality of Nvidia’s. Its share price has more than doubled this year.
Shares in US server maker Super Micro Computer, which is a crucial company in Nvidia’s supply chain, fell 5.5% on Friday after it identified “material weaknesses” in its internal control over financial reporting.
