DigestAI news desk

Cut through the AI noise.

Hardware & Compute3 min read

TSMC will reportedly supply 3nm base dies for Nvidia's NVHBM chips

Nvidia announced its NVHBM custom high‑bandwidth memory line in August 2026, aiming to standardise how HBM is built by moving its own memory controller into the base die. The company says the design could deliver up to a 30% bandwidth boost and 15% lower power consumption compared with standard HBM4E.

1 source

Key points

  • TSMC will reportedly provide its 3nm process for the base dies of Nvidia’s NVHBM memory stacks.
  • Samsung, SK Hynix and Micron are directed to use TSMC’s 3nm technology for NVHBM base dies, with Micron targeting volume production by 2027.
  • Nvidia claims NVHBM could boost memory bandwidth up to 30% and cut power consumption by 15% versus HBM4E.

According to the October 18 report, Taiwan Semiconductor Manufacturing Company will reportedly provide its 3nm process technology for the base dies of NVHBM stacks. Samsung Electronics, SK Hynix and Micron have been directed to use TSMC’s 3nm node for their NVHBM base dies, with Micron confirming plans to outsource both HBM4E and NVHBM base dies to TSMC and targeting volume production by 2027. Samsung’s involvement marks the first time it will work with TSMC on HBM base‑die manufacturing.

Full story from cryptobriefing.com · by Diego Almada Lopez · via Search: NVIDIAOpen source ↗

TSMC to build base dies for Nvidia’s custom HBM chips

cryptobriefing.com · 18 October 2026

Loading the full article…

This text was published by cryptobriefing.com and written by Diego Almada Lopez. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Hardware & Compute

All →

Related stories