Coughlin Associates and Objective Analysis released their 2023 report on emerging non-volatile memory technologies, which covers all aspects of new emerging non-volatile memories, who makes them, their applications and capital equipment requirements. The report has 272 page, 30 tables and 171 figures. Non-volatile memory technologies covered include various types of resistive randomw access memory (ReRAM), magnetic RAM (MRAM), ferro-electric RAM (FRAM) and phase change memory (PCM).
As the Internet of Things builds out, and as a growing number of devices make new measurements of data that was previously unmeasured, the world’s data processing needs will grow exponentially, including AI training and inference using data from these devices.
This growth will not be matched with increases in communication bandwidth, despite the adoption of new wireless standards like 5G.
In response to this mismatch, an increasing amount of processing will be performed at the edge, particularly in single-chip devices whose power consumption must be kept in check. This is the application that promises to benefit the most from today’s nascent adoption of emerging memory technologies.
IoT endpoints need to process and interpret this data, often in near real time. Combine this with the amount of data that must be stored and processed to train various types or artificial intelligence (AI) models, and the result is a great increase in demand for memory and storage.
Flash memory (NAND and NOR), DRAM, SRAM, and other popular memory technologies are facing potential technology limits to their continued improvement.
This has resulted in intense efforts to develop new memory technologies. Most of these new technologies are nonvolatile memories, which can be used for long-term storage or to provide a memory that retains information when powered down. Nonvolatile memories offer advantages for battery and ambient powered devices and also for energy savings in data centers.
Magnetic RAM (MRAM) and spin transfer torque RAM (STT MRAM) are starting to replace some NOR flash and SRAM and could possibly displace some DRAM within the next few years. The rate of development and increasing product volume in STT MRAM and other MRAM technologies will gradually result in lower prices.
The attractiveness of replacing volatile memory with high speed and high endurance nonvolatile memory make these technologies very competitive, assuming that their volume increases to reduce production costs (and thus purchase prices).
The use of a nonvolatile technology as an embedded memory combined with CMOS logic has great importance in the electronics industry. NOR flash in embedded devices reached its scaling limit at 28nm, and is being replaced with one of these new non-volatile technologies (especially MRAM and ReRAM).
As a replacement for a multi-transistor SRAM, MRAM could dramatically reduce the number of memory transistors and thus provide a low cost, higher-density solution. A number of enterprise and consumer devices currently use MRAM, to act as an embedded cache memory, and this trend will continue.
It is projected that total MRAM baseline annual shipping capacity will rise from an estimated 133TB in 2022 to 4.56EB in 2033. Total MRAM revenues are expected to increase from $118M in 2022 to about $98.3B by 2033 (as shown in the chart below compared to projected revenues for DRAM and NAND.
COUGHLIN ASSOCIATES AND OBJECTIVE ANALYSIS CHART
Much of this revenue gain will be at the expense of SRAM, NOR flash and some DRAM, although STT-RAM is developing its own special place in the pantheon of shipping memory technologies.
The demand for MRAM will drive demand for capital equipment to manufacture these devices. While MRAM can be built on standard CMOS circuits supplied by large semiconductor fabricators, MRAM and STT MRAM do require specialized fabrication equipment for the MRAM layers that is similar to or the same as that used in manufacturing the magnetic read sensors in hard disk drives.
The increasing demand for nonvolatile memory based upon MRAM will cause total manufacturing equipment revenue used for making the MRAM devices to rise from an estimated $55.1M in 2022 to between $302M to $4.3B by 2033 with a baseline projected spending of $2.2B.
A new report projects that MRAM and other emerging memory revenues could approach $1B and drive capital equipment revenues to $2.2B by 2033. These memories would replace embedded NOR and SRAM and possibly DRAM to support IoT and AI workflows.
https://www.forbes.com/sites/tomcoughlin/2023/09/06/emerging-non-volatile-memories-enable-iot-and-ai-growth/amp/
