TSMC: Nvidia’s AI GPU shortage to last 1.5 years

Rate this post

TSMC’s president acknowledged that the ongoing short supply of GPUs for artificial intelligence (AI) and high-performance computing (HPC) applications is due to limitations in its chip-on-wafer-on-substrate (CoWoS) packaging capabilities. The shortage is expected to persist for about 18 months due to increasing demand for generative AI applications and the relatively slow expansion of CoWoS capabilities at TSMC.

“It’s not a lack of AI chips, it’s a lack of our CoWoS capabilities,” TSMC president Mark Liu said in a conversation with Nikkei at Semicon Taiwan. “Currently, we cannot meet 100% of our customers’ needs, but we try to support about 80%. We think this is a temporary phenomenon. After the expansion of our (advanced chip packaging capacity), it should be reduced. Half a year .”

Leave a Comment