Investing.com-- Shares of NVIDIA’s (NASDAQ:NVDA) Asian suppliers fell on Thursday, tracking an overnight decline in U.S. tech amid growing doubts over a supply glut and limited demand for artificial intelligence infrastructure.
TSMC (TW:2330) (NYSE:TSM)- by far Nvidia’s biggest supplier- fell about 2% in Taipei trade, while peer Hon Hai Precision Industry Co Ltd (TW:2317), also known as Foxconn (SS:601138), fell more than 3%.
South Korea’s SK Hynix Inc (KS:000660), which supplies memory chips to Nvidia, fell about 3%, while Japan’s Advantest Corp. (TYO:6857)- which supplies chip testing equipment- slid more than 7%.
Losses in Asian markets came tracking a nearly 6% overnight decline in Nvidia, after TD Cowen said in a note that major AI investor Microsoft (NASDAQ:MSFT) was cancelling data center leases in the U.S. and Europe. The company had walked away from over two gigawatts of capacity in both the U.S. and Europe in the past six months, and had also canceled existing leases in the past month.
The TD note was a follow-up to a February note identifying data center lease cancellations by Microsoft through TD’s channel checks.
TD said the lease cancellations were linked to Microsoft’s decision to not support incremental AI training workloads for OpenAI. But the investment bank also flagged concerns over data center oversupply, given that Microsoft is one of the biggest investors in the fast-growing technology.
Still, TD noted that other AI hyperscalers Google (NASDAQ:GOOGL) and Meta (NASDAQ:META) were stepping in to backfill some of the capacity Microsoft had dropped, both in international markets and in the U.S.
The TD note comes just days after Alibaba (NYSE:BABA) Chair Joe Tsai warned of a data center bubble in the U.S., stating that several projects were being commissioned with no consideration for demand or customers.
Data center demand could also ease as the AI industry shifts towards inference- ie, processing and creating data, which is a less intensive process- from training AI models. But analysts have forecast that strong consumer demand for AI could help underpin demand for computing capacity, although this remains to be seen.