Could the AI-sparked clouds be cleared?

I look forward to seeing Samsung Electronics successfully mass-produce the 5th generation high-bandwidth memory (HBM3E) to be used for Nvidia’s high-end AI chips.

CHO WON-KYEONGThe author is a professor at UNIST and the head of the Global Industry-University Cooperation Center. Investors are deeply concerned about the sudden downturn in U.S. growth. The stock market crashed due to the fear of recession, but I am not convinced. In early August, stock prices plunged while bond prices soared. It could have been an offshoot of the rise in the yen’s value after the Bank of Japan’s rare rate hike and the following liquidation of the coveted yen carry trade referring to investors borrowing the yen at a low-interest rate and purchasing dollars or other currencies for high-interest rates on their bonds.

The AI algorithm trade is also considered a factor behind the crash of the New York stock market. Stock prices hovered at record highs thanks to a multitude of investors borrowing the cheap yen and buying tech stocks in the market. Just in time, Nvidia and big tech bubbles were mentioned. Amid the money movement, the news that Nvidia’s next generation AI chip (B100 GPU) would be delayed by more than three months due to its design defect fueled the market confusion.

This year, the key player in the historic stock market boom in the United States was AI for sure. The first half of the year effectively illustrated how AI beat the high interest rate. Nvidia CEO Jenson Huang said that AI and accelerated computing have opened a new industrial revolution. Could overly large AI investments by big tech lead to enhanced productivity? Skepticism arose that the total amount of capital expenditure on AI is excessive. Many insiders started to check if AI really can raise productivity.

There have been two bold investments in AI around the world before, though they were not comparable to the current investments. It is not likely that another AI winter will come and the boom in AI investment will cool down. How nice would it be if the unit labor cost goes down and the core inflation can be eased thanks to the enhanced productivity in the sectors that embraced AI early on? The Goldilocks economy of increased growth and stabilized prices won’t be possible right away. It is problematic that the global economy is overly dependent on Nvidia chips. Fortunately, Apple chose Google Cloud to pretrain the chips to be used in its AI system — Apple Intelligence. It is a desirable move for diversifying options. I look forward to seeing Samsung Electronics successfully mass-produce the 5th generation high-bandwidth memory (HBM3E) to be used for Nvidia’s high-end AI chips.