You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Investing.com-- SoftBank Corp (TYO:9434) said on Tuesday its unit Saimemory had formed a collaboration with Intel Corporation (NASDAQ:INTC) to develop a new class of memory chip technology.
The two will collaborate on commercializing “Z-Angle operation memory technology,” with the goal of creating high capacity, high bandwidth, and low power consumption chips.
The technology will be developed with the goal of being used in artificial intelligence data centers, Softbank said in a statement. High-speed memory is a key component of AI data centers, given the massive computational power required to run generative AI models.
Softbank Corp is a publicly listed unit of SoftBank Group Corp. (TYO:9984) and is also the conglomerate’s primary operating company. Softbank Group had in mid-2025 agreed to invest $2 billion in Intel.
Tuesday’s announcement comes as Intel seeks to shore up its chip offerings and catch up with its rivals in supplying to the AI sector.
Under the partnership, Saimemory will leverage Intel’s next generation DRAM bonding initiative and aim to create prototypes by at least early-2028, the companies said.
I've read three articles about ZAM today. The story has frightening similarities to Optane and PCMS. And, about a decade late, Intel is hiring a new chief GPU architect. Does LBT drive a DeLorean?
I've read three articles about ZAM today. The story has frightening similarities to Optane and PCMS. And, about a decade late, Intel is hiring a new chief GPU architect. Does LBT drive a DeLorean?
Funny, the guy who oversaw Optane just left Intel... as a VP. If they've learned anything they'll use Softbank like they got used by Micron: "Sure Masayoshi Son, throw 9 figures at it every year and let us know how it goes, we'll be ready."
I'm not a memory person, maybe Fred Chen can help us with this? I understand there is a lot of important technology inside Intel that should be let out but I do not get this one.
SoftBank does not have the ability to manufacture memory. Does that mean Intel is rejoining the race of memory manufacturing for the third time?
Sadly, due to financial pressure, market conditions, and timing, Intel couldn’t stay in the memory manufacturing business. Now, while NAND memory is in hot demand and short supply, Intel had already sold its NAND memory division to SK Hynix in 2020. What if …
SoftBank does not have the ability to manufacture memory. Does that mean Intel is rejoining the race of memory manufacturing for the third time?
Sadly, due to financial pressure, market conditions, and timing, Intel couldn’t stay in the memory manufacturing business. Now, while NAND memory is in hot demand and short supply, Intel had already sold its NAND memory division to SK Hynix in 2020. What if …
If Intel still had its memory business, it would be the highest revenue business unit at Intel. Intel will probably get back into memory right before the crash LOL
That said, from what I hear, this is not really a new memory type as much as a optimized DRAM for GPU accelerators. Intels model is to use if for internal products and offer it to other companies
If Intel still had its memory business, it would be the highest revenue business unit at Intel. Intel will probably get back into memory right before the crash LOL
Nonsense. Intel exited the DRAM business 40 years ago. Until four months ago, Micron stock was not an especially great investment over the long run, but, yeah, Intel should have lost money for 40 more years just so it could take advantage of the latest shorter-than-two-quarter DRAM and HBM pricing circus.
If Intel still had its memory business, it would be the highest revenue business unit at Intel. Intel will probably get back into memory right before the crash LOL
That said, from what I hear, this is not really a new memory type as much as a optimized DRAM for GPU accelerators. Intels model is to use if for internal products and offer it to other companies
Note: to be less confusing to my fellow blogges, here is my revised edition:
Yes, Intel missed this golden opportunity (and many other memory or non-memory opportunities over the years). I almost feel as if there is a curse placed on Intel.
Nanya Technology, a smaller memory manufacturer in Taiwan, has seen its one‑year stock performance jump almost tenfold, despite its technology being two to three generations behind Samsung, SK Hynix, and Micron.
Intel, on the other hand, sold its NAND division to SK Hynix in 2020 and completed the transaction in March 2025. Intel’s intellectual property, R&D employees, and manufacturing assets related to NAND were transferred to SK Hynix through this sale. While NAND prices and demand are rising sharply, additional NAND supply is limited. Intel has lost the opportunity to participate in this NAND boom.
Intel, on the other hand, sold its NAND division to SK Hynix in 2020 and completed the transaction in March 2025. Intel’s intellectual property, R&D employees, and manufacturing assets related to NAND were transferred to SK Hynix through this sale. While NAND prices and demand are rising sharply, additional NAND supply is limited. Intel has lost the opportunity to participate in this NAND boom.
So let me understand this edited version. You are thinking that Intel was showing a lack of foresight about AI software using NAND storage in 2020/21, when NAND prices were falling rapidly, and Intel wanted to exit the SSD business, which was really intended for Optane. You can't be serious.
NAND SSDs had/have a lower margin than Intel's chip businesses, so the NAND business was a drag on Intel's margins. I'm unaware of any 2025 actions, but the NAND business was spun out as a separate company, Solidigm, in 2021, became a subsidiary of SK Hynix, and the original CEO, Rob Crooke, was a former Intel SVP who led the storage group. (I know Rob, and IMO he's one of the best senior executives I met at Intel.)
Intel has/had a lot of faults, but not a seeing a 2H2025 NAND opportunity in 2021 is not one of them.
I'm not a memory person, maybe Fred Chen can help us with this? I understand there is a lot of important technology inside Intel that should be let out but I do not get this one.
