Two German companies want to commercialize DRAM+ together. This brand-new non-volatile alternative to classic DRAM can serve a niche, theoretically making production in Europe financially viable.
Neumoda and Ferroelectric Memory (FMC) are collaborating on the development, testing, and eventual sale of non-volatile DRAM memory under the name DRAM+. The two German companies have the explicit ambition to bring not only the development but also the production of memory to German factories with this new technology.
The core of this plan is the DRAM+ memory. This memory integrates a new technology based on ferroelectric hafnium oxide (HfO2). The HfO2 replaces the classic capacitor in DRAM with an alternative that is non-volatile, thus retaining data during power loss. Neumoda and FMC integrate the technology with DRAM to create a form of memory that retains data (like SSDs) but is still super fast (like classic RAM).
Innovation for Capacity
The idea is not new: so-called FeRAM technology with a ferroelectric layer has existed for a while but traditionally struggles with capacity issues. Existing technology only supported a few megabytes of memory. The two German companies innovate by embracing HfO2. This material is scalable and can be integrated into the classic DRAM production process. In principle, capacities of a gigabyte are possible.
With such specifications, DRAM+ memory can be relevant for numerous applications. Neumoda and FMC are thinking of applications within AI, but also in the medical and industrial sectors.
read also
DRAM and NAND to Become More Expensive Worldwide
DRAM+ would have interesting niche applications that could justify production in Germany. Building traditional memory at scale is not advantageous in Europe and mainly happens in factories of Samsung and Micron. DRAM+ targets a separate niche that could justify local fabrication. Through their own development and patents, this is precisely what the two parties are aiming for.
On the other hand, non-volatile memory always generates enthusiasm in the planning phases but has never really broken through. In 2019, Intel developed 3DXpoint (Optane) memory together with Micron, which was fast but non-volatile. Optane was supposed to sit between SSD and RAM and fulfill a relevant niche role there. That sounded good on paper, but the relationship between the added value, complexity, and price was not right, so the technology was discontinued after a few years.