Get your free copy of Editor’s Digest
FT editor Roula Khalaf picks her favourite stories in this weekly newsletter.
When will artificial intelligence start to replace human workers? This is a question that has generated much speculation amid the AI ​​boom. But long before we worry about that happening, a shortage of human workers may become the biggest obstacle for the AI ​​industry.
Taiwan Semiconductor Manufacturing Co. (TSMC), the world’s largest chipmaker and maker of the chips powering the AI ​​revolution, saw its sales rise 45% to $7.9 billion in July, helping drive strong second-quarter growth. Demand for AI chips remains strong, and its high-performance computing business accounted for more than half of TSMC’s sales last quarter.
Despite these impressive numbers, AI stocks have been volatile in recent months as investors have grown increasingly wary of risks to companies like TSMC, ranging from the Taiwan earthquake to Sino-Taiwan tensions to broader geopolitical stresses. Less discussed, but just as important, is the looming crisis of a shortage of engineers and technicians.
Until now, the prevailing view has been that expanding semiconductor manufacturing capacity is simply a matter of money. The global semiconductor shortage that began in early 2020 was addressed by governments pouring billions of dollars into semiconductor manufacturers to boost production capacity, preferably in their own factories. TSMC has been expanding semiconductor factories in the United States, Germany and Japan.
The United States is one of the world’s most aggressive producers of semiconductor capacity, with investment in the industry expected to reach more than $250 billion over the next five years, but it’s turning out that cash alone can only sustain so much in chip manufacturing.
The problem is that building a chip factory isn’t as easy as setting up a new factory in another country to assemble smartphones and quickly hiring and training local workers. Running a chip factory requires highly skilled employees with master’s and doctoral degrees in science and engineering. Building the chip manufacturing plants themselves also requires specialized workers.
Major investment and subsequent expansion in the U.S. semiconductor industry will create more than 160,000 new jobs in engineering and technician support, as well as related construction and technical employment. McKinsey analysisBut only about 1,500 engineers join the chip industry each year. For chip technicians, the numbers are even lower, with about 1,000 new technicians joining the industry each year. Over the next five years, demand for these workers is projected to reach 75,000.
Meanwhile, the U.S. semiconductor manufacturing workforce is down 43% from its peak in 2000, according to McKinsey. At the current pace, the shortage of engineers and technicians could reach 146,000 by 2029. In South Korea, home to chipmaker Samsung Electronics, the semiconductor industry has been struggling with a labor shortage since 2022 and is expected to face a labor shortfall of 56,000 by 2031, according to industry estimates.
Demographics are also an issue. Taiwan and South Korea, where TSMC and Samsung base most of their employees, are facing declining populations; the number of students enrolled in higher education has fallen every year since 2012. The two countries account for more than 80% of the world’s contract chipmaking. Labor shortages are already becoming more severe. Pushed back TSMC’s Arizona factory opened on Nov. 15, 2017, and about half of the plant’s 2,200 employees are reportedly flying in from Taiwan. Cultural differences further complicate recruitment.
Building a new factory would cost nearly $30 billion, and to commercially justify that price tag, the factory would need to operate nonstop, 24/7. TSMC founder Morris Chang said: pointed out In the United States, if a machine breaks down at 1 a.m., it will be fixed the next morning, but in Taiwan, it will be fixed at 2 a.m. This Taiwanese work culture may be difficult to replicate in other countries.
So can AI start manufacturing chips? Indeed, AI is helping design, test, and validate new designs, accelerating the development of new chips. Producing physical chips from those designs is an entirely different story. The need for experienced engineers to operate machines is unlikely to be solved by AI in the near future.
Companies are rightly struggling to fill jobs that require high skills and qualifications, but in the semiconductor industry, the gap between jobs and workers is widening dangerously.
june.yoon@ft.com
