As the global stock market continues to thrive amid a surge in artificial intelligence (AI) investments, experts predict that this upward trend may persist for some time. They advise focusing on the semiconductor and power sectors, which are benefiting from AI hardware value chains.
Lee Jin-young, head of the Premier Family Office at Shinhan Bank's Banpo Center, stated on July 9 at the 'ABC Launch Commemoration AI Ecosystem Innovation Forum' held at The Plaza Hotel in Seoul, "The market is currently entering the latter stages of a bull market, and we are beginning to see the formation of an AI bubble. Historically, bubbles have lasted an average of five years, and considering the interest rate cycle, this bull market has at least one more year to go."
Lee noted that the current wave of technological innovation is fostering widespread optimism, making it unlikely for the AI investment cycle to reverse quickly. He explained, "If data center investments pause for just six months, competitors will secure more semiconductors and greater AI computing power during that time. Taking a break from investment for one to two years would be akin to turning previously invested assets into sunk costs, essentially abandoning the platform business."
He added, "Like the 'prisoner's dilemma' in economics, big tech companies are in a position where they cannot afford to stop investing, as they are constantly watching each other. The competition for AI investment is likely to continue."
Lee anticipates that AI hardware companies will directly benefit more from this investment competition than big tech firms. He remarked, "The funds used by big tech will ultimately flow to memory companies like Samsung Electronics and SK Hynix, as well as GPU, CPU, and power infrastructure firms. While hardware companies in East Asia are experiencing significant increases in earnings per share (EPS), the profit growth rate for hyperscalers is relatively limited."
As the AI investment cycle evolves, the sectors benefiting from it are also changing. Initially, NVIDIA's GPUs and high-bandwidth memory (HBM) led AI training, but the market is now shifting towards inference, increasing the importance of CPUs, DRAM, and NAND flash.
Lee analyzed that since AI requires physical infrastructure like data centers, new beneficiary industries emerge whenever bottlenecks occur, not just in GPUs and memory but also in networking equipment and power facilities. He predicted, "In the future, the expansion of sovereign AI and physical AI will broaden the related value chain even further."
He views the recent adjustments in semiconductor stocks as more indicative of short-term cyclical trading rather than a downturn in the AI investment cycle. Lee stated, "While big tech stocks have shown renewed strength, the key question is where big tech is allocating its funds. In terms of stock prices, the AI hardware value chain is likely to be more favorable."
He noted, "Samsung Electronics' price-to-earnings ratio (PER) has dropped to 4.9, and Micron's has fallen to 7. Despite continuous growth in performance, valuations are decreasing, making it difficult to assert that 'the bubble has already burst' or 'the bubble has ended.'"
However, he cautioned that the AI investment cycle will not last indefinitely and highlighted potential warning signs. He identified a surge in margin loans, indiscriminate initial public offerings (IPOs), speculative overheating centered on unprofitable companies, and aggressive interest rate hikes by the Federal Reserve as key indicators of market peaks. Additionally, he advised that the AI investment and funding trends of hyperscalers like Meta and Oracle will serve as important leading indicators for the sustainability of the AI investment cycle.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.