Global semiconductor shares took a sharp dive on Tuesday following a blunt assessment from Broadcom CEO Hock Tan, who likened the current wave of artificial intelligence investment to gambling in a casino. Speaking at a conference in Singapore, Tan warned that many investors are pouring capital into the sector without a fundamental understanding of the technology or the risks involved. He argued that unrealistic expectations have created a bubble where people are essentially betting on outcomes they cannot predict, rather than basing decisions on sustainable business models.
The fallout was felt most acutely across Asian markets, where indices plummeted as traders offloaded chip stocks. South Korea’s Kospi index crashed over six percent, with industry giants Samsung Electronics and SK Hynix seeing significant losses. This downward trend extended to Japan and Taiwan, dragging down key players like Tokyo Electron and TSMC. Even in the United States, futures indicated a shaky start for Wall Street, with heavyweights including Nvidia and Micron feeling the pressure of shifting sentiment.
Adding fuel to the fire was news regarding Chinese memory chip maker CXMT, which reportedly made breakthroughs in its DRAM architecture. The prospect of increased memory density from a rising Chinese competitor sparked fears among shareholders that established leaders like Micron and Samsung could lose their competitive edge. These technical developments combined with Tan’s warnings highlighted a growing anxiety that the massive debts being taken on to fund AI infrastructure may not yield immediate returns.
Tan specifically pointed toward high profile ventures like OpenAI to illustrate his point about profitability. While noting that huge annualized revenues suggest growth potential, he reminded listeners that these companies continue to bleed cash due to astronomical training costs. Using a vivid analogy involving children spending tokens at a video arcade, Tan cautioned that early corporate interest often manifests as superficial pilot projects rather than deep integration, leaving some to wonder if the current AI gold rush is built on solid ground or mere speculation.
