Longsys Unveils Innovative Storage Solutions for Edge AI at FMS 2026

At the Future of Memory and Storage (FMS) 2026 conference in San Jose, California, which took place on August 5, Chief Scientist Dr. Jian Chen from Longsys introduced the company's pioneering approach to edge AI storage. During his keynote speech titled "Intelligence at the Edge Powered by the Storage Foundry Model," Dr. Chen discussed the significant shift of AI processing from centralized cloud systems to decentralized edge environments, highlighting the rising demand for specialized storage solutions.

The Storage Foundry Model

Longsys has developed the Storage Foundry Model, a framework designed to innovate edge AI storage solutions. This model includes advanced technologies such as HLCache™, iSA™ + AISSD™, and AIDIMM™, which are particularly engineered to tackle prevalent industry challenges, including soaring DRAM prices, inadequate memory availability, and bandwidth limitations. By addressing these issues, Longsys aims to create new pathways for innovation in edge AI technologies.

Dr. Chen emphasized the notion of 'collaboration' in developing these solutions. He explained how the fragmented nature of the edge AI market, which he likens to a "thousand-handed challenge," necessitates a more collaborative and innovative approach rather than reliance on one-size-fits-all storage products.

Key Innovations Presented

1. HLCache™

HLCache™ leverages High-Level Cache technology to effectively manage the flow of data between UFS and DRAM. This results in diminished memory strain and enhanced multitasking performance in mobile AI devices, allowing for smoother operations and improved user experiences.

2. iSA™ + AISSD™

This innovative pairing facilitates efficient local AI inference for devices such as AI PCs and AI agent hosts. By offloading certain operational parameters and cache data to dedicated SSDs, these solutions reduce DRAM reliance while still accommodating extensive AI models, pushing the boundaries of what mobile devices can achieve.

3. AIDIMM™

AIDIMM™ takes a revolutionary approach to memory management in AI Box and edge systems. With a native 256-bit interface capable of achieving bandwidths up to 307.2GB/s, it offers a high-bandwidth memory architecture that simplifies hardware upgrades while minimizing redesign efforts.

A Call for Collaboration: The 'Collabovate' Philosophy

To fortify the manufacturing and development landscape for edge AI technologies, Dr. Chen introduced the concept of 'Collabovate,' a blend of collaboration and innovation. With this initiative, Longsys invites industry partners to join in creating a versatile ecosystem that fosters advanced edge AI solutions. By combining expertise across various sectors including chips, devices, and AI models, Longsys aims to streamline the development and deployment processes, ultimately enhancing performance and reducing barriers for edge AI technologies.

Conclusion

Longsys's vision for the future challenges traditional views on storage solutions. By adopting a comprehensive model that encourages cross-domain cooperation and tailored innovations, the company is positioning itself as a leader in the edge AI space. Dr. Chen's address at FMS 2026 not only highlighted the technological advancements Longsys brings to the table but also set the foundation for collaborative efforts to evolve the industry's approach to edge AI storage solutions. As the demand for intelligent computing accelerates, the Storage Foundry Model may very well redefine success in this burgeoning field.

For more information about Longsys and its range of storage solutions, visit their official website at Longsys and stay updated through their social media channels.