Emerging memories are poised for another growth surge.
That’s according to the annual report released jointly authored by Objective Analysis and Coughlin Associates. It’s projecting emerging memories to be a $44 billion market by 2031 by displacing incumbent technologies including NOR flash, SRAM, and DRAM, either in the form of standalone memory chips and embedded memories within microcontrollers, ASICs, and even compute processors.
Over time, the authors expect emerging memories to create new markets of their own, and that’s there’s a great deal of competitive advantage to be gained for participating in the market—not just for memory makers and foundries, but also designers and users of SoCs who are already incorporating these new non-volatile memories into their designs to achieve much more competitive power consumption and system responsiveness.
click for full size image
MRAM and ReRAM are forecast to replace the bulk of embedded NOR and SRAM in SoCs, and even displace DRAM and NAND flash for some applications. (Courtesy: Objective Analysis)
The revenue growth around emerging memories will be driven in part by the need for new tooling to support different materials and processes, and this will provide a growth boost for the capital equipment market. For example, the report expects total MRAM manufacturing equipment revenue will grow to more than one-hundred times its $10 million 2020 total to reach $1.1 billion in 2031, while standalone MRAM and STT-RAM revenues will grow to about $1.7 billion, or over forty-two times 2020’s standalone MRAM revenues. Along with embedded ReRAM, embedded MRAM will compete to replace the bulk of embedded NOR and SRAM in SoCs.
From a sheer dollars perspective, 3D XPoint memory could reach revenues of more $20 billion by 2031, the report predicts, which echoes last year’s report that it would help to drive PCRAM uptake. However, it also illustrates where some of the challenges may be for emerging memories.
Right now, Intel’s Optane is the only commercially 3D Xpoint memory available, and it requires new software to fully exploit it, said Jim Handy, principal analyst with Objective Analysis. In addition, the target customers most likely to take advantage of are in the financial sector. “They tend to be enormously conservative.” And although 3D Xpoint is a PCRAM-based technology, Optane’s fate is very much tied to Intel Xeon processors, he said, especially now that has exited the Optane SSD business due to lack of adoption.
What makes Optane DIMMs appealing is that they’re cheaper than DRAM, said Handy, about half as much per gigabyte, according to Intel. So for what you spend on DRAM in a system, you could get twice as much Optane. He said it’s good for the customer, but not necessarily good for Intel because it’s still struggling to make Optane profitably.
click for full size image
Standalone Memory Annual Petabyte Shipments, 2019-2031 (Courtesy: Objective Analysis / Coughlin Associates)
Making these emerging memories profitably has always been a barrier to their wider adoption. The immediate opportunities for growing the MRAM and ReRAM markets are beneath the covers within an SoC as NOR flash and SRAM hit their limitations, said Handy. MRAM-based SoCs are already shipping, and ReRAM SoCs are on the verge of coming out. “This this year is a pretty good indication of how things will change in the future.”
As noted in the report, the appeal and adoption of STT MRAM is due to its compatibility of MRAM and STT-RAM processes with conventional CMOS processes because these memories can be built directly on top of CMOS logic wafers. This makes MRAM a viable NOR Flash and SRAM replacement, as flash memory doesn’t have the same compatibility with conventional CMOS, and the power savings of non-volatile and simpler MRAM and STT MRAM when compared with SRAM is significant.
Despite the potential market growth for MRAM, not all the players are going to survive to enjoy the spoils. The assents of Spin Memory are being liquidated and its patents likely to be sold, said Handy. The shuttering of Spin Memory isn’t necessarily a reflection of the health of the overall MRAM market, but that the company’s revenue in the short term wasn’t enough to finance their business. That the company opted to build their own backend fab to put MRAM on top of a piece of standard CMS logic was expensive and probably made it unattractive to investors, he said.
But it’s just as much the foundries as memory makers that are pushing MRAM forward, such as TSMC, Samsung and GlobalFoundries, Handy said. “They’re the ones who have the advanced process technologies.” That’s what’s going to lead to MRAM replacing SRAM and NOR Flash because it will allow chips to shrink with the logic. He said at a certain point that will bring MRAM prices down low enough to be adopted by other applications that were reluctant because it was too expensive.
In addition to looking at PCM, ReRAM, FRAM and MRAM, the annual report on emerging memories also looks at the competitive strengths and weaknesses of less mainstream technologies and includes company profiles encompassing chip makers, technology licensors, foundries, production tool makers, and research consortia.
>> This article was originally published on our sister site, EE Times.
- Dealing with the memory problem in artificial intelligence
- Is nonvolatile MRAM right for your consumer embedded device application?
- Flash 101: NAND Flash vs NOR Flash
- Emerging memory technologies ready to break out
- Edge AI challenges memory technology
For more Embedded, subscribe to Embedded’s weekly email newsletter.