Intel is now looking to capitalize on the ongoing DRAM demand by partnering with a SoftBank subsidiary to launch a new "ZAM" memory technology.
Intel's ZAM Memory Modules Now Bring In Higher Power Efficiency Through Innovative Interconnect & EMIB
Well, with the AI infrastructure buildout being at its peak this year, there has been a gigantic rise in DRAM demand, driven by the adoption of hyperscalers and chip manufacturers. And more importantly, given that memory suppliers are limited worldwide, the supply chain bottleneck is immense right now, underscoring the need for new competitors to enter, which is why Intel is reported to be taking a new angle with the memory business. It is claimed that Team Blue will partner with SoftBank's Saimemory to develop a new standard called Z-Angle Memory (ZAM).
It is reported that efforts to develop the ZAM memory technology had commenced under the Advanced Memory Technology (AMT) program initiated by the DoE, during which Intel showcased its "next-generation" DRAM bonding. Well, SoftBank's press release doesn't specifically discuss how the Z-Angle Memory is expected to be positioned. Still, based on what we know about Intel's DRAM bonding, ZAM would likely feature a staggered interconnect topology routing connections diagonally within the die stack rather than drilling straight down.
Standard memory architectures cannot meet the needs of artificial intelligence. NGDB defines a completely new approach that will accelerate our move into the next decade.
We are rethinking how DRAM is organized to fundamentally advance computer system architecture, aiming to achieve orders-of-magnitude performance improvements and incorporate innovation into industry standards.
- Intel officials on next-gen DRAM bonding
By the "Z-Angle" approach, Intel will effectively utilize a significant portion of the silicon area for memory cells, allowing higher densities and lower thermal resistance. Based on how the DRAM bonding technology could look, we could see ZAM using a copper-to-copper hybrid bonding, which enables more efficient inter-layer fusion and creates a "monolithic-like" silicon block rather than separate stacks. And since it is said that ZAM will be a capacitor-less design, we will see Intel's EMIB used to connect the memory to the AI chip.
SoftBank's collaboration with Intel on ZAM ultimately allows it to own the memory stack as well, which we could see debut with the company's custom ASICs, such as the Izanagi lineup, giving the firm greater control over the architectural layout. While figures on the effective uplifts of ZAM versus HBM aren't specific for now, the Z-Angle approach provides the memory with better power efficiency and higher densities, ultimately enabling higher-layer stacking. Here's how HBM vs ZAM could play out:
- 40-50% Lower Power Consumption
- Simplified Manufacturing Through Z-Angle Interconnects
- Higher Storage Per Chip (Up to 512 GB)
This isn't the first time Intel has pursued something in the DRAM segment; the company had a dedicated business around it, but it exited it in 1985 due to a drop in market share following competition from Japanese vendors. However, now that memory has created such a significant opportunity for companies, it would be interesting to see whether Intel's ZAM can make an impact, and one effective way to do so is by convincing leaders like NVIDIA to integrate the technology.
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
