SoftBank’s Saimemory and Intel Forge Strategic Partnership to Power Next-Gen AI Memory Revolution
In a move that signals intensifying competition in AI hardware innovation, SoftBank’s Tokyo-based subsidiary SAIMEMORY has teamed up with Intel Corporation to co-develop and commercialize Z-Angle Memory (ZAM) — a next-generation memory architecture designed for the rising demands of artificial intelligence workloads. (ソフトバンク)
This collaboration, announced on February 2, 2026, aims to address one of the fastest-growing bottlenecks in AI computing: memory capacity, speed, and energy efficiency. Current high-performance AI models — particularly those used in data centers for training and inference — require memory that can keep pace with explosive growth in data throughput while limiting power consumption and cost. Z-Angle Memory is SoftBank and Intel’s answer to this challenge. (ソフトバンク)
Under the agreement, SAIMEMORY will lead the development and commercialization efforts, drawing on Intel’s technical expertise in advanced DRAM and stacking technologies, particularly Intel’s Next Generation DRAM Bonding (NGDB) initiative — a foundational research program supported by the U.S. Department of Energy’s Advanced Memory Technology (AMT) initiative. The partnership intends to produce working prototypes by fiscal 2027 (ending March 31, 2028) and achieve commercial availability by fiscal 2029. (ソフトバンク)
Why This Matters Now
As AI models scale to tens or even hundreds of billions of parameters, traditional memory systems — including High-Bandwidth Memory (HBM) — struggle to deliver the required capacity, bandwidth, and energy efficiency. Z-Angle Memory promises to bridge this gap by offering high capacity, enhanced bandwidth, and significantly reduced power consumption, which could dramatically improve performance and operating costs for AI-centric data centers and cloud services. (ソフトバンク)
The industry has taken notice: markets reacted favorably to the announcement, with reports of Intel’s shares climbing on the news of the collaboration, illustrating investor confidence in AI hardware innovation as a growth driver for both companies. (TechStock²)
SoftBank’s Strategic AI Playbook
Establishing SAIMEMORY as a dedicated R&D hub — alongside collaborations with global semiconductor leaders like Intel — is part of SoftBank’s broader strategy to bolster its footprint in the AI ecosystem. The company is already active in data center infrastructure and AI acceleration partnerships with other tech giants (including NVIDIA) and continues to invest heavily in hardware innovation to capture more value from the AI transformation sweeping through technology and enterprise sectors. (NVIDIA Newsroom)
Implications and Outlook
If successful, Z-Angle Memory could shift the memory technology landscape by offering alternatives to legacy DRAM and HBM paradigms, enabling data centers to run larger AI workloads with less energy and greater throughput. For Intel, the tie-up strengthens its position in AI-oriented semiconductors; for SoftBank, it reinforces a long-term commitment to building AI-ready infrastructure and chip ecosystems that underpin next-generation computing. (ソフトバンク)
Glossary
-
DRAM (Dynamic Random-Access Memory): A type of volatile memory widely used in computers and servers, essential for fast data access but traditionally limited by power and bandwidth constraints.
-
High-Bandwidth Memory (HBM): A high-speed memory interface for graphics and AI accelerators that stacks memory dies to provide greater bandwidth but is more costly and complex to manufacture.
-
Memory Bandwidth: The rate at which data can be read from or written to memory — critical for high-speed AI computations.
-
AI Training vs. Inference: Training is the process of teaching an AI model using vast datasets; inference is using the trained model to make predictions or decisions in real-time applications.
Source: https://www.techinasia.com/news/softbank-unit-intel-partner-on-next-gen-memory-for-ai (ソフトバンク)
