Could RAM replace SSD? Well, not really but this plucky Japanese startup may come very close

The memory-storage pyramid may end up becoming less complicated with super cheap RAM

When you purchase through links on our site, we may earn an affiliate commission.Here’s how it works.

A Japanese memory startup that has been lurking in the shadows for more than a decade has just emerged with a bang. Neo Semiconductor announced that it wants to produce DRAM chips that are 8x denser than the current crop - 16Gbit DRAM - using a technology called 3D X-DRAM.

As the name hints to, there will be layers of material, 230 in total for now, helping it produce a 128Gbit DRAM chip. In comparison,Samsungis aiming to launch a 32Gbit DDR5 DRAM chip in 2023 with 1TB memory modules on the horizon.

According to data revealed by the co-founder and CEO of the company, Andy Hsu, the first prototypes are expected to land next year should conversations to license the technology to DRAM manufacturers (Micron, Samsung Semi, SK Hynix, Kingston Technology) come to fruition.

Similarly to 3D NAND, 3D X-DRAM should enable memory densities to increase exponentially reaching up to 1Tb before 2024. In comparison, it took the DRAM industry more than a decade to move from 4Gb to 16Gb memory chips, a tiny improvement in comparison.

Cheaper and denser

Cheaper and denser

Memory modules are relatively cheap at the lower end of the spectrum and become atrociously expensive at the very top end. A single 256GB DDR4 serverRAMmodule, the maximum memory size on the market, retails for around $2,500 (or $10 per GB). In comparison, you can get 32GB RAM for less than $60 (or less than $2 per GB).

Neo’s technology solution could bring down the cost of memory as dramatically as 3D NAND did for solid state storage; imagine 256GB of RAM at $60. What makes it even more enticing is that it uses existing manufacturing techniques to achieve the layering, similar to 3D NAND.

This comes at a time where server manufacturers are finding it very difficult to add more memory slots to their server motherboards, forcing them to adopt exotic solutions likeCXLexpansion cards. 3D X-DRAM would also help solve this by boosting memory density.

Are you a pro? Subscribe to our newsletter

Are you a pro? Subscribe to our newsletter

Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!

Popular applications like machine learning or AI (thinkChatGPT) that useLLM(Large language models) require access to very large pools of memory and this comes at a significant cost, both financially and in terms of power/latency.

The RAM drive

While I don’t expect it to reach end users anytime soon, it has the potential to change the memory-storage pyramid asIntel3D XPoint (AKAOptane) failed to impose itself as a candidate tier between system memory andSSD.

For end users, given the fact that system storage has reached a plateau (most laptops come with 256GB onboard storage these days), one can expect to see devices that run on RAM only in a remote future. A paradigm shift that could change the overall computing landscape and the way modernoperating systemsoperate.

Without the need for “slow” SSD and access to a single pool of ultra fast system memory, applications would be faster and potentially more secure. After all,VPN providerslikeNordVPNorExpressVPNhave rolled outRAM-only serversthat load a new system image after every reboot.

Désiré has been musing and writing about technology during a career spanning four decades. He dabbled inwebsite buildersandweb hostingwhen DHTML and frames were in vogue and started narrating about the impact of technology on society just before the start of the Y2K hysteria at the turn of the last millennium.

Google puts Nvidia on high alert as it showcases Trillium, its rival AI chip, while promising to bring H200 Tensor Core GPUs within days

A new form of macOS malware is being used by devious North Korean hackers

This new malware utilizes a rare programming language to evade traditional detection methods