About Broadcom, what the market argues about is never demand—it’s supply. The AI business of $AVGOB —XPU chips custom-built for a few of the largest cloud providers—has orders already queued up to the absolute production capacity limit. The real bottleneck is not something you can immediately expand by just throwing money at it: it’s the advanced packaging stage. This leads to a counterintuitive conclusion: during an AI boom cycle, supply constraints are actually a good thing. With such constrained supply, order visibility is extremely high, and revenue outlook for the next few quarters is clearer than for most chip companies. But the other side of the coin is margin pressure and cash-flow stress: expanding capacity requires capital, and customer-specific customization discounting is also eroding profits. Existing analysis has pointed out that this is its biggest risk going forward. My view is that this model is a sweet kind of frustration in a bull market. But once market sentiment weakens, the same facts will be reframed just as quickly as a “growth ceiling.” When the stock price starts to drop then, it won’t be gentle. Which side are you on: treating the capacity bottleneck as a moat, or as a hidden risk? Share your reasoning.