SoftBank Corp. Unit Teams Up With Intel to Work on Next-Generation Memory Tech — Update

By Yang Jie and Kosaku Narioka

A unit of SoftBank. Corp is teaming up with Intel to develop cutting-edge memory technology that can better support artificial intelligence.

The Japanese telecom company said Tuesday that subsidiary Saimemory will work on the technology with chip maker Intel, aiming to bring it to market in around 2029.

According to people familiar with the matter, SoftBank plans to invest up to 3 billion yen, or about $19 million, over the next two years to advance the project, though the overall research-and-development cost has yet to be finalized.

SoftBank said in a statement that the collaboration will focus on developing memory that can support high‑capacity data processing and reduce power consumption in data centers used for training and running large-scale AI models.

The pairing is strategic, coming as AI drives up memory costs and energy bills for data centers increasingly seen as key tech for national economic security. But analysts say it will take considerable time before a newcomer can pose any real challenge to the "Big Three" of memory makers: Samsung Electronics and SK Hynix of South Korea, and U.S. company Micron.

For Intel and SoftBank to truly shake things up, they would likely need to partner with a current AI‑chip leader like Nvidia, said Jukan Choe, an analyst with Citrini Research.

The announcement comes as memory-chip companies grapple with a surge in demand from fast‑growing AI firms that is squeezing already tight supplies.

The market strain has intensified governments' efforts to secure more resilient chip supply chains, with countries like Japan and the U.S. stepping up cooperation on advanced manufacturing.

Write to Yang Jie at jie.yang@wsj.com and Kosaku Narioka at kosaku.narioka@wsj.com

(END) Dow Jones Newswires

February 03, 2026 02:14 ET (07:14 GMT)