Chip Stocks Slide After Broadcom CEO Compares AI Investing to 'Casino'

Semiconductor shares fell on Tuesday after Broadcom chief executive Hock Tan compared many AI investors to gamblers. Chinese memory chip maker CXMT also said CXMT had made good progress on a new chip architecture. South Korea's benchmark Kospi index fell 6.25% as investors sold chip stocks. Samsung Electronics shares dropped 7.4% on the South Korean exchange. Memory chip maker SK Hynix shares fell 10%. Japan's Nikkei index fell 1.8%. Taiwan's stock exchange lost almost 2%, and Taiwan chip maker TSMC fell by a similar amount. Japan's chip equipment maker Tokyo Electron fell 7.15%. SoftBank Group, an OpenAI investor, lost 8.5%. US stock futures pointed to a mixed opening on Wall Street after the technology-heavy Nasdaq composite slipped 0.4% on Monday. US chip maker Micron Technology fell 3.9%. Nvidia, Advanced Micro Devices and Broadcom also fell. Crude oil prices dropped amid escalating tensions in the Middle East. AI shares have fluctuated sharply in recent months amid concerns about large, debt-funded investments in AI infrastructure. The latest sell-off followed a Monday South China Morning Post report that CXMT had made good progress on new DRAM architecture that would increase memory density. Investors worry that the technology could challenge Samsung, SK Hynix and Micron, the world's biggest memory chip makers. CXMT debuted on the Shanghai stock market in late July. CXMT makes DRAM chips, which store the data that other AI chips use for calculations. Bloomberg reported Tan's comments at a conference in Singapore. Tan said the AI boom had attracted investors who did not really understand their investments. Tan said many investors set extreme expectations before discovering that investing was "almost like they are in a casino". AI companies such as ChatGPT maker OpenAI remain loss-making. Tan said OpenAI's annualised revenues of $50 billion (£38 billion) at the end of September suggested a viable business model but offered little guarantee of long-term profitability. Tan said OpenAI was still losing money because OpenAI was spending heavily on training. Tan said the revenue figure demonstrated potential revenue growth from providing frontier generative AI models rather than establishing profitability. Tan noted that a sizeable portion of AI laboratories' enterprise revenue came from pilot projects. Tan compared such projects to a child's visit to a video arcade: "Your parents give you $10 of tokens, you finish it in 10 minutes and you haven't learned anything."