SoftBank’s Saimemory and Intel launch Z-Angle Memory to build AI-ready DRAM

SoftBank unit teams with Intel to chase AI-ready DRAM

Z-Angle Memory partnership aims to plug AI DRAM gap

Intel is partnering with Saimemory, a SoftBank subsidiary, to commercialize next-generation DRAM under the Z-Angle Memory (ZAM) program, aiming for prototypes by the fiscal year ending March 31, 2028 and full commercialization in fiscal 2029. The collaboration leverages Intel’s work in the U.S. Department of Energy’s Advanced Memory Technology program and seeks to address surging memory demands from artificial intelligence and high-performance computing workloads that current architectures struggle to serve.

Intel says the ZAM effort combines a new memory architecture and assembly approach that improves DRAM performance while lowering power consumption and cost, with an explicit emphasis on energy efficiency as AI computing drives rising data‑center power use. Dr. Joshua Fryman, Intel Fellow and CTO of Intel Government Technologies, says “standard memory architectures aren’t meeting AI needs,” and frames the joint program as a route to scale designs that can be validated across data‑center workloads and AI models to reduce total cost of ownership for operators.

Saimemory, established in December 2024, will draw on Intel’s IP, manufacturing know‑how and joint testing facilities to accelerate prototype iterations, with reports that Fujitsu and other partners are also involved in large‑scale validation. The partners position ZAM as a potential lever to relieve acute supply constraints in the global memory chain caused by AI‑driven demand outpacing supply; successful scaling through fiscal 2029 could influence DRAM roadmaps and competitive dynamics among memory suppliers.

Intel’s push into AI memory comes amid broader strategic moves

Separately, Intel signals a deeper push into AI compute by appointing a new chief architect to lead GPU development, CEO Lip-Bu Tan says at the Cisco AI Summit, while declining to name the hire. Tan notes it takes “some persuasion” to attract top GPU talent in a market dominated by specialist rivals, underscoring the challenge Intel faces as it seeks to compete across the AI compute stack.

The company is also navigating production snags and supply issues that recently overshadow its quarterly results, and analysts are watching for named anchor customers and clearer foundry commitments as Intel scales GPU designs, refines yields and seeks to convert investment momentum into durable revenue growth.