For nearly four decades, Taiwan Semiconductor Manufacturing Co. (TSMC) adhered to a strict rule: build the world’s best logic processors and leave memory manufacturing to SK Hynix, Samsung, and Micron. In November 2026, the relentless memory hunger of frontier AI has pushed TSMC to reconsider that boundary.
According to supply-chain reports from Taipei, TSMC has formed an executive planning group led by Advanced Packaging Vice President Hsu Kuo-chin to evaluate entering advanced 3D DRAM production — a move directly requested by Nvidia as traditional High Bandwidth Memory (HBM) supply struggles to keep pace with GPU compute scaling.
Why Nvidia Pushed TSMC Toward 3D DRAM
- TSMC has initiated preliminary planning to enter advanced memory manufacturing — specifically vertically stacked 3D DRAM — at the direct urging of its largest AI customer, Nvidia.
- The internal task force is led by TSMC Advanced Packaging VP Hsu Kuo-chin and is evaluating whether to construct dedicated memory fab lines or acquire/convert existing capacity.
- The move blurs the 35-year boundary between pure-play logic foundries (TSMC) and the "Big Three" memory makers (SK Hynix, Samsung, and Micron).
“In post-2027 AI architectures, memory is no longer a component sitting beside the GPU — it is bonded directly onto the logic wafer.”—
- ✓Sources: written from primary sources and official statements, linked above.
- ✓Fact-check: names, dates and claims checked against the original sources.
- ✓AI use: researched with AI tools; written and reviewed by our editors. Our AI policy
- ✓Independence: UpdateNext has no financial relationship with the companies covered.



