FMS 2026 | Longsys Chief Scientist Jian Chen Highlights the Storage Foundry Model for Edge AI

PR Newswire

SAN JOSE, Calif., Aug. 7, 2026 /PRNewswire/ — At FMS 2026 on August 5, Longsys (301308.SZ) Chief Scientist Dr. Jian Chen delivered a keynote speech titled “Intelligence at the Edge: Powered by the Storage Foundry Model,” addressing the new storage demands created as AI shifts from the cloud to the edge.

Intelligence at the Edge: Powered by the Storage Foundry Model

During the keynote, Dr. Chen introduced Longsys’ Storage Foundry Model and presented three edge AI storage innovations — HLCache™, iSA™ + AISSD™, and AIDIMM™ — designed to address key industry challenges including high DRAM costs, limited memory capacity, and bandwidth constraints, creating a new pathway for edge AI storage innovation.

Engineering Culture: The Foundation of Storage Innovation

Dr. Chen traced Longsys’ growth since its founding in 1999, including the launch of FORESEE, the acquisition of Lexar, expanded packaging and manufacturing through Longforce (Suzhou) Technology, and the integration of Brazil-based Zilia. Longsys now operates a global manufacturing network supporting worldwide delivery.

Rather than focusing solely on standardized products like traditional storage providers, Longsys applies full-stack engineering across eight areas — chip design, firmware development, material engineering, host software and OS-level adaptation, hardware design, packaging, automated testing, and mass production — enabling cross-domain hardware-software co-optimization and customized development. Together, these integrated capabilities across the storage value chain form the basis of the Storage Foundry Model.

Longsys Storage Foundry Model

Redefining Storage Collaboration for the Edge AI Era

Dr. Chen highlighted that while NAND Flash and DRAM have achieved significant cost reductions over decades, the rise of edge AI has introduced a new level of complexity. He described the fragmented edge AI landscape as a “thousand-handed challenge,” where diverse architectures, AI models, device forms, and applications require coordinated solutions rather than standardized approaches.

Thousand-handed challenge

From AI PCs and mobile devices to industrial systems and robotics, different scenarios require tailored trade-offs among performance, power efficiency, form factor, and security. The Storage Foundry Model goes beyond conventional storage offerings by enabling solutions tailored to customers’ CPUs, SoCs, AI models, and system architectures — integrating hardware, firmware, and system-level optimization to accelerate development cycles and lower barriers to edge AI deployment.

Three Longsys Innovations Powering Edge AI

HLCache™ UFS leverages Longsys’ self-developed High-Level Cache technology to intelligently manage hot and cold data between UFS and DRAM, reducing memory pressure and improving multitasking performance on mobile AI devices.

iSA™ + AISSD™ enables efficient local AI inference for AI PCs and AI agent hosts by offloading MoE parameters and KV cache to dedicated SSD resources, reducing DRAM usage while supporting large-scale AI models.

AIDIMM™ addresses memory bandwidth challenges in AI BOX and edge AI systems with a plug-in high-bandwidth memory architecture, featuring a native 256-bit interface and up to 307.2GB/s bandwidth. Its flexible design simplifies system upgrades and reduces hardware redesign efforts.

AIDIMM

“Collabovate”: Building the Future of Edge AI Together

Dr. Chen highlighted that as edge AI evolves from early adoption to large-scale deployment, diverse requirements across devices, architectures, and workloads cannot be addressed by a single product or technology.

To address this fragmented market, Dr. Chen introduced “Collabovate” — combining Collaboration and Innovation — as a new ecosystem collaboration approach. Through the Storage Foundry Model, Longsys is bringing together partners across chips, devices, and AI ecosystems to leverage complementary capabilities and accelerate customized edge AI solutions.

By opening its full-stack technology capabilities, Longsys aims to build a more collaborative storage ecosystem, enabling deeper industry collaboration and unlocking new possibilities for intelligent computing.

About Longsys

Founded in 1999, Longsys (301308.SZ) is a globally leading branded semiconductor memory enterprise, integrating R&D, design, packaging and testing, manufacturing, and sales services. For more information please visit https://www.longsys.com/, and follow Longsys on LinkedIn and Facebook.

Cision View original content to download multimedia:https://www.prnewswire.com/news-releases/fms-2026–longsys-chief-scientist-jian-chen-highlights-the-storage-foundry-model-for-edge-ai-302846008.html

SOURCE Longsys

About The Author