Now Playing
TSMC Terafab Talks, AMD On-Device AI & China's DUV Stockpile
Audio is available on Spreaker — see link below.
TSMC enters confirmed Terafab partnership discussions as AMD runs a 320B parameter model locally on Ryzen AI hardware — reshaping assumptions about where AI workloads actually run. Plus China's $13B DUV stockpile, TSMC's $164B US expansion signal, and what Tesla's memory cuts mean for DRAM demand.
Audio is available on Spreaker — see link below.
TSMC is now in early confirmed discussions with Elon Musk's Terafab project for a potential Texas fab partnership. That's not a rumour anymore.
Set against that uncertainty, TSMC's broader US commitment is not ambiguous at all. The company raised its twenty twenty six capex guidance to sixty to sixty four billion dollars and separately announced an additional hundred billion dollar commitment to Arizona.
While TSMC expands westward, China has been building in a different direction. Chinese fabs accumulated three hundred and forty three immersion DUV lithography machines by early twenty twenty six.
The other major development worth tracking is AMD's demonstration of a three hundred and twenty billion parameter model running locally on the Ryzen AI Max Plus PRO four ninety five. The model ran at twenty tokens per second on-device.
Two shorter signals round out today's picture. Tesla halved the LPDDR five memory in its AI five chip and cut AI six LPDDR six by one third.
Two things to keep close. First, the Terafab ownership structure.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.