Taiwan AI firms warn of new supply chain bottlenecks amid demand boom

Taipei, June 4 (CNA) Taiwanese companies across the AI supply chain said Thursday they expect strong growth ahead as AI investment accelerates, but warned that bottlenecks are emerging in areas ranging from memory and advanced packaging to semiconductor materials analysis.

At a Taiwan Stock Exchange (TWSE) media briefing held alongside Computex Taipei, executives from chip designer Alchip Technologies, memory maker Nanya Technology, semiconductor analysis provider MSSCORPS Co. and substrate supplier Unimicron Technology outlined the challenges facing the industry's next stage of expansion.

One emerging issue is semiconductor materials analysis, which MSSCORPS said will become increasingly important as chipmakers move toward more advanced process technologies.

"Materials are becoming the next scaling bottleneck," said Gene Liu (柳淳浩), CEO of MSSCORPS's U.S. subsidiary, arguing that the shift to angstrom-era processes and AI-driven computing is increasing demand for advanced characterization technologies, proprietary analytical expertise and expanded laboratory capacity.

The comments came as AI-related demand continues to grow across Taiwan's semiconductor ecosystem.

Nanya Technology Senior Vice President Joseph Wu (吳志祥) said AI demand is expected to support annual industrywide DRAM bit growth of more than 20 percent in the coming years.

To capitalize on that trend, Nanya plans to raise research and development spending by up to 70 percent and double its output within two to three years through capacity expansion, Wu said.

He said Nanya recently completed a US$2.5 billion private placement and now has about US$5.8 billion in cash, which will support its future expansion plans.

At Unimicron Technology, Director Victor Hsu (徐興源) said AI-related business is expected to account for more than 60 percent of the company's revenue this year.

Hsu said the rapid upgrade cycle for AI chips and servers is driving demand for advanced ABF substrates and packaging materials, while increasing pressure on manufacturers to expand capacity.

Meanwhile, AI-related projects now account for about 80 percent of Alchip Technologies' revenue, Chief Financial Officer Daniel Wang (王德善) said.

Wang said that reflects growing demand from cloud service providers for custom AI chips designed to improve efficiency and reduce infrastructure costs.

He added that the slowdown in Moore's Law has made advanced packaging technologies such as 2.5D packaging and CoWoS increasingly important, creating new capacity constraints across the industry.

The event was part of TWSE's efforts to showcase Taiwan's semiconductor ecosystem during Computex.

TWSE said it expects about 40 IPO applicants this year, around 40 percent of which will be AI-related.