The company selling the shovels just suggested the gold miners might be gambling. Broadcom CEO Hock Tan, whose firm booked $16.7 billion in AI revenue last quarter, compared many AI investors to casino gamblers at the DBS Asia Future Forward Conference on October 12, 2026.
Markets took the remark personally. The iShares Semiconductor ETF (SOXX) fell around 3% that same day, and Broadcom’s own shares slipped about 2%.
Not everyone accepted the premise. Aoifinn Devitt of Moneta pushed back, arguing that the current AI market correction is a natural one and that the casino comparison overstates the risks investors are actually taking on.
What Tan actually said
Tan also likened enterprise spending on AI tokens to kids at an arcade, burning through tokens quickly and walking away having learned nothing.
Tan also aimed some skepticism at the AI labs themselves. Many of these labs lose money, and he questioned whether that model holds up over the long run, even with healthy revenue coming in now.
He pointed to OpenAI as an example of strong near-term numbers. According to Tan, the company’s annualized revenue run rate reached approximately $50 billion by September 2026.
Perhaps his most consequential comment was about money itself. Tan said access to capital may become the main limit on how fast AI infrastructure can grow, more so than the supply of chips or electricity.
The numbers behind the warning
Broadcom reported $16.7 billion in AI semiconductor revenue for fiscal Q3. That figure was up 221% from a year earlier.
Forecasts point to $21.7 billion in AI revenue for Q4 and a projected $58 billion for fiscal 2026.
Broadcom’s own long-range outlook is even bigger. The company anticipates its annual AI chip revenue could reach $230 billion by fiscal 2028.
The one-day reaction was modest but telling. A roughly 3% drop in a broad semiconductor ETF suggests the market heard Tan’s words as a comment on the whole sector, not just a quirk of one executive’s speaking style.
Devitt’s counterpoint
Devitt’s argument reframes the same facts in a calmer light. Her objection centers on the casino analogy specifically. Gambling implies odds stacked against the player and outcomes driven by chance, and Devitt contends that this description exaggerates the actual risks facing AI investors.
What this means for AI and chip investors
The most important thread in Tan’s remarks may be the capital point, not the casino line. If funding becomes the bottleneck, then Broadcom’s growth depends on investors continuing to bankroll AI labs, including the loss-making ones Tan questioned.
For investors, the practical takeaway is to separate the layers of the AI trade. A supplier reporting $16.7 billion in quarterly AI revenue sits in a very different position from a lab that is growing fast but still losing money.
The next test is Broadcom’s Q4 report. Results near the forecast $21.7 billion would suggest demand is holding up despite the CEO’s caution, while a miss would give the casino analogy more weight than Devitt thinks it deserves.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.