Nvidia’s AI-powered video cards will remain a rarity until at least the end of 2024. This is what TSMC President Mark Liu said in an interview with Nikkei. Liu blames the lack of GPU on the lack of on-chip on-substrate chip packaging capacity.

TSMC expects the shortage to continue for another eighteen months, as demand for AI GPUs continues to rise, while TSMC’s CoWoS packaging technology capacity is very slowly expanded. Louie informed Nikki Capacity is expected to be expanded enough within a year and a half to be able to meet full demand. Currently, TSMC says it can meet about eighty percent of demand.

TSMC is the only company that makes Nvidia H100 and A100 AI data center GPUs. This device is used to run many AI applications, including ChatGPT. Like most AI chips, the two GPUs use Nvidia CoWoS. The company claims that demand for products using this advanced packaging technology rose unexpectedly last year and is now nearly three times higher than the same period last year. TSMC says it is installing new tools at its existing packaging plants to boost capacity, but this will take time.