archy13 - stock.adobe.com
Micron Technology will immediately cease development of 3D XPoint memory and shift resources to products based on the emerging Compute Express Link standard for connecting compute, memory and storage.
The memory and storage manufacturer, based in Boise, Idaho, determined there was "insufficient market validation to justify the ongoing high levels of investments required to successfully commercialize 3D XPoint at scale to address the evolving memory and storage needs of its customers," according to a press release. Micron said that, although development would end now, the manufacturing of 3D XPoint products would not wind down until it completes industry commitments over the next several quarters.
Micron began jointly developing 3D XPoint memory with Intel in 2012 to fill the performance gap between more expensive DRAM and cheaper NAND flash. In 2019, Micron bought out Intel's interest in the IM Flash Technologies fabrication plant that is dedicated to 3D XPoint production in Lehi, Utah, and in March 2020, they ended their joint development program. Although Micron and Intel ended their joint development work, Micron had continued to supply 3D XPoint wafers to Intel under a contractual agreement.
Micron's president and CEO Sanjay Mehrotra said during a conference call that the company hopes to finalize the sale of the Lehi fab by the end of 2021 and is in discussions with several potential buyers.
In response to the news, Intel issued the following statement: "Micron's announcement doesn't change our strategy for Intel Optane or our ability to supply Intel Optane products to our customers."
Jim Handy, general director and semiconductor analyst at Objective Analysis, said Intel likely had put together a contingency plan, since Micron has slowly been extricating itself from the 3D XPoint business. He said he expects Intel to continue to promote and keep 3D XPoint technology as a competitive advantage over AMD processors. Handy noted that Intel could acquire the Lehi fab or ramp up its Rio Rancho, N.M., manufacturing site that runs prototype lots.
Jim HandyGeneral director and semiconductor analyst, Objective Analysis
Micron's decision was more sudden than Handy anticipated it would be, but he said he could understand the rationale.
3D XPoint market has been sluggish
"The market for 3D XPoint has grown significantly more slowly than Intel and Micron had hoped back in 2015 and earlier when they committed to the effort," Handy said. "This entire time, both companies have been losing money. Since the SSD market didn't develop, and since that was supposed to be the volume driver that would help them drive production costs down, it was getting hard to see how the product could become profitable in the foreseeable future."
Joseph Unsworth, a research vice president at Gartner, said he wasn't surprised to see Micron end its 3D XPoint work, given that the company had yet to commercialize any SSD or other products in volume and "must have been losing considerable money on the technology." But Unsworth did not think the Micron decision will kill 3D XPoint. "It certainly complicates the supply/manufacturing situation and puts Intel as a true single-source," Unsworth said.
Greg Wong, founder and principal analyst at Forward Insights, noted that "one of the major concerns for customers" is that there is no dual source for Optane, and "Micron's decision kills any hope of a dual source." Wong said he thought Micron would give 3D XPoint a couple more years before deciding whether to continue.
Micron said it plans to apply the knowledge gained from its 3D XPoint development, as well as related engineering expertise and resources, to new types of memory products that can address the requirements of AI and data analytics workloads. The company said it sees great promise in new classes of memory-centric technologies that use Compute Express Link (CXL) to scale capacity, performance and content.
Raj Hazra, senior vice president and general manager of the compute and networking business unit at Micron, told TechTarget earlier this year that CXL would "open the door to an unprecedented level of system architecture innovations starting in the next 18 to 24 months." He said CXL would facilitate different ways to connect memory and storage with compute.
"It's an open industry specification, so it allows more players to innovate on a standard bus," Hazra added. "It's no longer a proprietary interface that you have to license for a particular CPU. So, whether you're building an accelerator, storage class memory or storage devices, the CXL standard gives you the ability to put your innovation on that compute platform."
Carol Sliwa has been a TechTarget senior writer since 2008. Her coverage area includes enterprise architecture, flash, memory and storage drive technology.