South Korean chip stocks fell sharply on Wednesday as rising bond yields and growing concerns over the scale and returns of Big Tech’s artificial intelligence spending triggered a wider selloff in semiconductor shares. The Kospi dropped as much as 6.8 per cent before recovering some ground, while Samsung Electronics and SK Hynix each fell more than 8 per cent at one point. The decline followed a sharp fall in US technology and semiconductor stocks a day earlier, with investors now questioning whether the huge spending behind the AI boom can continue if borrowing costs remain high.

The pressure spread across Asia. A Bloomberg gauge tracking Asian semiconductor stocks fell around 3.2 per cent, while Japan’s Kioxia Holdings dropped as much as 11 per cent and Taiwan Semiconductor Manufacturing Company fell nearly 2 per cent. The immediate trigger was the rise in US Treasury yields. The 30-year US Treasury yield touched 5.34 per cent on Tuesday, its highest level since 2007, while the 10-year yield moved above 4.7 per cent. Higher yields generally make future corporate earnings less valuable in present terms, putting particular pressure on high-growth technology companies whose valuations depend heavily on expectations of stronger profits several years ahead.

For the AI industry, the concern goes beyond share valuations. Major technology companies are committing enormous amounts of money to data centres, servers, networking equipment and advanced chips, with some of that spending increasingly supported through debt. A Reuters analysis had noted that borrowing by large AI-focused technology companies, often described as “hyperscalers”, has become a factor in the bond market. A Wall Street Journal analysis has also estimated that nine major technology companies have around $3 trillion in off-balance-sheet commitments, much of it connected to AI infrastructure. Investors are therefore asking whether the expected returns from AI will be large enough to justify the cost of building that infrastructure if financing becomes more expensive.

The semiconductor industry is particularly exposed because chipmakers sit at the centre of the AI infrastructure cycle. Samsung Electronics and SK Hynix are major suppliers of memory chips used in AI servers, while Taiwan Semiconductor Manufacturing Company produces advanced processors for some of the world’s largest technology companies. Their shares had benefited strongly from the AI investment boom, leaving them vulnerable when investors began cutting exposure to high-growth technology stocks. The Philadelphia Semiconductor Index fell around 5 per cent on Tuesday, while the Nasdaq Composite dropped 1.33 per cent, with Nvidia, Micron Technology and other AI-linked companies also coming under pressure.

Higher oil prices are adding to the uncertainty. Brent crude moved above $91 a barrel after rising for a fourth consecutive session, while West Texas Intermediate traded around $85. Rising energy costs can make it harder for inflation to cool, potentially keeping central banks cautious about cutting interest rates. That creates another problem for technology companies because elevated rates increase borrowing costs at a time when AI infrastructure spending remains exceptionally high. Investors are also watching the Federal Reserve for clues about inflation, interest rates and the economy, while longer-term Treasury yields continue to rise because of factors including government borrowing, bond issuance and energy prices.

South Korea is particularly sensitive to the semiconductor downturn because chipmakers account for a significant portion of its equity market. The latest fall therefore reflects more than routine profit-taking. It shows that investors are becoming increasingly focused on whether the AI investment cycle can deliver sufficient returns to justify the hundreds of billions being committed to it. If bond yields remain elevated, the pressure could spread beyond semiconductor companies to the wider technology sector. For now, Samsung and SK Hynix remain at the centre of that repricing, with markets weighing strong long-term demand for AI hardware against a more expensive funding environment.