Filtered to AI Hardware & Semiconductors · show all articles
Nvidia never sold you a GPU. Your last laptop or phone purchase still got more expensive because of one — and the mechanism is simpler than it sounds.
Morris Chang founded TSMC on a bet the rest of the industry considered dubious: a chipmaker that designs nothing of its own. It became the single structural precondition for the fabless industry this cohort spends most of its pages covering.
This cohort has profiled ASML, Applied Materials, Lam Research, KLA, and Tokyo Electron individually. Here they are side by side — the complete map of who builds the equipment behind every chip in this cohort.
"Export controls" gets discussed constantly as geopolitics. It is, first and mechanically, a specific list of tools and chips that may not cross a specific border — and the mechanism matters more than the politics for understanding what actually happens next.
Gate-all-around transistors and backside power delivery bought the industry another decade. The deeper problem underneath — Dennard scaling's collapse and doubling fab costs — is the one nobody has a clean fix for yet.
"Foundry" gets used constantly in AI hardware coverage without ever being explained. The concept it names is simpler than it sounds, and its history explains almost everything else in this cohort's foundry briefings.
Taiwan's second foundry sits a few kilometers from TSMC's headquarters and competes for almost none of the same customers — a deliberate positioning that a 93% revenue jump in its core node suggests is working.
Nvidia designs. AMD designs. Apple designs. Almost none of them manufacture. One Taiwanese company makes the leading-edge chips for nearly everyone who doesn't, and that single fact shapes more of the AI boom than any GPU announcement.
Every AI accelerator headline is about the die. The actual ceiling on how many of them the world can build in 2026 sits one manufacturing step later, in a packaging process most coverage never names.
Building a leading-edge fab outside Taiwan was long treated as economically irrational. TSMC is doing it anyway, at a scale that has quadrupled since the first announcement — and the reasons are as much about customers as about geopolitics.