C.C. Wei, the CEO of TSMC, has issued a warning that global chips will continue to lag behind the surging demand driven by AI for several years.

Bloomberg cited Wei’s comments during the chip foundry’s annual shareholders meeting this week that, despite forecasting sales growth of more than 30% for this year, long-term supply faces a shortfall.

Insatiable demand for AI hardware has been compounded by a dearth of memory hardware, with leading suppliers SK Hynix, Samsung, and Micron stretching lead times out toward the end of the decade.

The situation in the Strait of Hormuz has only exacerbated the AI hardware crisis, severing supplies of critical single-source process chemicals required for advanced chip fabrication, most notably, helium. The Taiwanese giant, which manufactures chips for Nvidia, AMD, and Broadcom, said in March it was monitoring the supply chain situation related to helium.

In a bid to bolster its manufacturing capacity – and appease the Trump administration following the threat of tariffs – TSMC is working to add new manufacturing capacity in Arizona.

But Bloomberg reported Wei telling shareholders that even with the new capacity, TSMC won’t be able to fulfill demand stemming from U.S. customers.

“It will be a long time before we can meet customer demand,” Wei was reported as saying.

Leaders from Hewlett Packard Enterprise and Arista have already sounded the alarm bells over the memory shortage and its potential knock-on impact on infrastructure buildouts.

For the memory makers, meanwhile, the shortage has proved a financial boon, with two of the big three memory makers witnessing their market caps pass the $1 trillion mark amid unprecedented demand.

SK Hynix previously said that its lead times were being pushed out as far as 2028, with plans to build low-power wafer factories to address the supply chain issue.