What Legacy Will Money-Losing Optane Have?

Article By : Gary Hilson

Adding a tier of memory between DRAM and NAND is now widely seen as a worthy goal.

After several years of promoting Optane, its version of 3D XPoint, as an enabling layer in the storage/memory hierarchy between NAND flash below and DRAM above, Intel last month whispered in its Q2 earnings call that it was abandoning its Optane technology.

“I almost missed it,” Jim Handy, principal analyst with Objective Analysis, said of the notation.

Micron Technology’s March 2021 announcement that it would not be pursuing 3D XPoint technology was less hushed, which was unsurprising. Outside of an SSD shared with close partners, the company didn’t ship any commercial 3D XPoint products in volume.

But while Intel abandoning Optane may have surprised many, there were signs that the technology’s days were numbered, said Handy, as the company had stopped talking about Optane for the past year. This is not without precedent and usually a signal that Intel is poised to phase out a product line. “Optane certainly was being given the silent treatment for a long time,” he added.

And Optane did not make money for Intel either.

Handy said Intel’s sale of its NAND business unit to SK Hynix meant that Optane was moved over to the company’s data-center business, making it more obvious as to how much of a loss Intel was incurring with Optane. “They were losing money hand over fist.”

Intel could justify the losses because if it makes a more profitable processor, it can use that profit to pay for Optane losses, Handy added. “If anybody else makes Optane, they’re just going to lose money.”

If no one had to worry about making a profit, 3D XPoint technology would naturally become a part of the storage/memory hierarchy, he said. But no matter the performance of the memory technology, it must be cost-effective.

Along with fellow analyst Tom Coughlin of Coughlin Associates, Handy has published an annual report of emerging memories, and their modeling showed that Optane would struggle to achieve the economies of scale needed to succeed. “Technical features are nice, but cheap always wins.”

So with Intel winding down Optane and Micron already out of the 3D XPoint business, what does it mean for that extra layer in the memory storage hierarchy, and what does it mean for the future of phase-change memory (PCM)?

3D XPOINT HAD A HEADY BEGINNING

When Intel and Micron jointly announced 3D XPoint in July 2015, there was much speculation as to what kind of memory it was. Most bets were being placed on PCM, also known as PCRAM, a technology grouped together with other emerging memories, including resistive random-access memory (ReRAM), magnetoresistive random access memory (MRAM), and ferroelectric random-access memory (FRAM).

The only commercialized PCRAM to date is in the form of Intel Optane DIMMs and SSDs, which the company has positioned as another layer in the storage hierarchy above 3D NAND SSDs and below DRAM. (Source: Intel Corporation)

At the time, Intel and Micron claimed 3D XPoint was a new class of memory, the first since the introduction of NAND flash, and that it offered non-volatile memory speeds up to 1,000× faster while being 8× to 10× times denser than DRAM. Few details were disclosed about the material system or switching mechanism, with the companies saying only that the switching mechanism is via changes in resistance of the bulk material and that the companies had “invented unique material compounds” to create the 3D XPoint memory.

While some coverage pegged 3D XPoint as being “bulk switching” ReRAM, today, Optane has been placed in the PCM category. By 2019, Handy and Coughlin’s emerging memories report had forecasted the revenue growth of 3D XPoint at $16 billion by 2029 in the form of Optane, driven in large part by its sub-DRAM prices. At that time, the technology jointly developed by Intel and Micron was the only PCM that had achieved any significant commercialization.

PCM has been the subject of research by some significant players, including IBM Research; the Advanced Memory Lab at Leti; and Intermolecular Inc., a part of the Performance Materials Business of Merck KGaA in Darmstadt, Germany, which operates as EMD Performance Materials in the U.S. and Canada. The key challenge facing PCM has been that it still doesn’t scale as well as the manufacturing technologies for vertical 3D NAND because it must be built up layer by layer—meaning a 128-layer PCM isn’t cost-competitive against existing technologies. Adding to that, the small dimensions create challenges because the materials involved all have unique deposition issues.

When Optane was released, there was excitement in the industry because it appeared as though Intel and Micron had addressed the key challenges sufficiently and that they could release a viable product. As with 3D NAND, which had a period where the manufacturing processes needed to be fine-tuned to make it economically feasible if it were to replace planar NAND, it looked as though what 3D XPoint needed was an ecosystem to support its adoption and a shaking out of the manufacturing process.

ECOSYSTEM VENDORS SAW OPTANE POTENTIAL

The ecosystem to support the adoption of Optane required both hardware and software.

With hardware alone, Intel allowed native access to the Optane persistent memory with app direct mode, but it required a new API and rewriting applications that had been written for DRAM. This prompted companies like memVerge to develop software to replicate what the hardware does in memory mode by employing tiering algorithms between various types of memory. By providing a software-defined DRAM-compatible interface to the applications, it negated the need for customers to rewrite their applications to leverage Optane.

In the meantime, Intermolecular announced it had developed deposition technology for 3D vertical memory arrays that could potentially shape the second iteration of 3D XPoint technology by using atomic-layer deposition (ALD) chalcogenides in lieu of physical-vapor deposition (PVD). The company’s device overcame the inability to stack tens of layers in a 3D structure, which limits memory density and results in higher costs. The prospect of making 3D XPoint technology less expensive came at a time when both the SSD and DIMM form factor of Optane were only just starting to see some volume sales.

Intermolecular’s ALD process allows for future 3D vertical integration with higher density and reduced cost, which could have paved the way for the expansion of the 3D XPoint market. (Source: Intermolecular Inc.)

An early Optane adopter was Lenovo, which introduced 11 products sporting Intel Optane in the DIMM form factor as part of an overall server and storage appliance refresh. Intel itself, meanwhile, combined its 144-layer QLC 3D NAND with Optane in an M.2 SSD, with Optane enabling instant on and responsiveness capabilities so that users could search and find files faster and start applications quicker. Optane could be placed in front of QLC NAND SSDs to collect data in bigger chunks, so it was optimized before landing on the QLC, which helped to solve the bottlenecking. This allowed for data to be laid out more efficiently—and extend the life of the QLC SSD.

The Intel Optane Memory H20 for clients combines 3D Xpoint technology with the company’s 144-layer QLC 3D NAND in the M.2 form factor. (Source: Intel Corporation)

As recently as a few months ago, it appeared as though Optane had a future, even if Intel wasn’t talking it up.

Research firm Yole Group’s “Emerging Non-Volatile Memory Report 2022” found that PCM had become a significant portion of the small sliver of pie that was made up of emerging memory, and it was thanks to Intel’s Optane. Its report found that PCM, MRAM, and ReRAM account for only 0.4% of the standalone memory market—NAND flash and DRAM still dominated, with 96.7% of the overall market.

Yole’s forecast for a 3D XPoint-driven PCM segment was that it would grow at a slow pace, and it assumed that Intel remained the only player involved in the commercialization of 3D XPoint, with the market penetration rate of its Optane products not expected to evolve significantly in the next two years.

PCM BEYOND OPTANE

With Intel’s shuttering of its Optane business, is the PCM market dead?

Not necessarily, said Handy, because Intel has processed a lot of PCM wafers due to its Optane efforts. The question, he said, remains: Will the knowledge gained from that work (or Micron’s, for that matter) ever see the light of day—to be applied elsewhere?

What will continue to persist is the notion that there’s a gap in the storage/memory hierarchy between DRAM and NAND worth addressing, and Intel’s got everybody thinking about adding a tier of memory, Handy added. “There might be something that comes of that.”

This article was originally published on EE Times.

Gary Hilson is a general contributing editor with a focus on memory and flash technologies for EE Times.

 

Virtual Event - PowerUP Asia 2024 is coming (May 21-23, 2024)

Power Semiconductor Innovations Toward Green Goals, Decarbonization and Sustainability

Day 1: GaN and SiC Semiconductors

Day 2: Power Semiconductors in Low- and High-Power Applications

Day 3: Power Semiconductor Packaging Technologies and Renewable Energy

Register to watch 30+ conference speeches and visit booths, download technical whitepapers.

Subscribe to Newsletter

Leave a comment