Semiconductor manufacturing plant clean room
Coughlin Associates and Objective Analysis released their 2024 report on evolving volatile memories, An In-Depth Look at New Memories. These memories include magnetic random access memory, MRAM; resistant random access memory ReRAM; ferroelectric random access memory, FRAM; as well as phase shift memory, PCM. The future of these developing memories depends on technical characteristics, developments in existing higher volume memories as well as emerging economies of scale.
In this market, economies of scale reign supreme. We saw this with Intel ending its Optane 3D XPoint campaign in 2022, and we had seen the same phenomenon play out before with NAND flash, whose high volume drove its costs below that of DRAM. The report delves deep into the impact of economies of scale and concludes that the success of NAND flash and the failure of Optane proved that wafer volume must approach 10% of the volume of a competing technology in order to achieve cost parity.
The first widespread use of emerging memory is for memory integrated on a CMOS logic chip, replacing NOR flash, which has reached its scalability limit at 28 nm and is today often being replaced by MRAM and ReRAM. Single-transistor MRAM cells are now competing with multi-transistor static random access memory, SRAM, to dramatically reduce the number of memory transistors on a chip in order to provide a lower-cost, high-density solution higher. Some enterprise, industrial and consumer devices currently use MRAM as an integrated memory and this trend will continue.
MRAM processes have already been developed on conventional CMOS logic processes, allowing them to be built directly on top of CMOS logic wafers, using fewer additional mask steps than a more conventional flash memory. Using an evolving non-volatile memory can provide significant power savings compared to SRAM. As the emerging memory cost per gigabyte ($/GB) approaches that of SRAM, this replacement could lead to significant market expansion.
Based on these economies of scale considerations, the report predicts that the total annual shipping capacity of core developing memory will grow from about 340 TB in 2023 to 8.46 EB in 2034. Total core developing memory revenue is expected to grow from $421 million in 2023 to about $71.7 billion by 2034. Much of this rapid growth in income will supported by the emerging memory shift of SRAM, NOR flash and some DRAM.
The evolving demand for memory should drive increasing demand for the capital equipment needed to manufacture these devices. This report models this demand by focusing on the MRAM market. While MRAM can be built on standard CMOS circuits supplied by major semiconductor fabricators, specialized fabrication equipment is required for the MRAM layers.
This accessory is similar to or the same as that used in the manufacture of magnetic readout sensors on disk drives. Other emerging memory technologies will require their own complement of additional hardware, much of it identical to tools that are already in widespread use.
The growing demand for new memory technologies will increase the total revenue of manufacturing equipment by an estimated $53.4 million in 2023, between $434 million to $3.8 billion by 2034 with a projected base spending of $2.4 billion . The chart below shows basic projections of annual production equipment costs.
Baseline forecasts for annual costs of memory manufacturing equipment in development
The new non-volatile memory report projects that the industry’s core revenue could grow to $71.7 billion by 2034 as it achieves economies of scale and displaces SRAM, NOR flash and some DRAM.