Samsung to use TSMC technology for Nvidia's next-generation memory chips, Seoul newspaper reports

Samsung is turning to direct competitor TSMC to manufacture critical components for Nvidia's next-generation artificial intelligence memory chips. In an unexpected reversal for its standalone semiconductor manufacturing business, Samsung accepted Nvidia's supply terms and agreed to outsource AI memory base dies to TSMC after internal foundry yields fell short of requirements. The production shift is not confined to Samsung alone.

Nvidia has instructed suppliers across the broader memory sector to tap TSMC's process technology, aligning Samsung, Micron, and Hynix to utilize TSMC-produced base dies for NVHBM hardware. While Reuters notes that Samsung's adoption of TSMC technology surfaced via a Seoul newspaper report, accounts differ on the primary driver: some sources focus on Samsung's internal manufacturing yield deficits, whereas others describe a top-down mandate dictated directly by Nvidia across all participating vendors. Crucial operational details are missing across the available coverage.

Published reports do not reveal production targets, delivery schedules, financial arrangements, or the precise technical generation of the upcoming memory chips. It is likewise unspecified how long Samsung, Micron, and Hynix will depend on TSMC base dies, or if Samsung intends to return production to its own foundries if internal yields recover.