FADU Unveils First File-Semantic Solid-State Drive
The new storage technology allows GPUs to access files directly, bypassing traditional CPU-bound routing methods.
Updated on Oct. 15, 2026 in Semiconductors
Live Poll
Do you believe new AI storage technologies will effectively reduce national energy consumption and costs?
FADU unveiled its file-semantic solid-state drive at the 2026 Open Compute Project Global Summit in San Jose. The technology manages files directly by using the FC6161 controller to bypass traditional block address systems.
Why it matters
By enabling GPUs and DPUs to process file requests without CPU intervention, this system significantly reduces computing overhead, power consumption, and latency in AI infrastructure.
The 16-terabyte drive utilizes the sixth-generation FC6161 controller to translate file-to-address requests directly. The system achieves a 2.5 times performance boost over Gen5 technology while maintaining a 28-watt peak power draw.
The players
FADU
FADU is a semiconductor company specializing in flash storage solutions and enterprise controller technology.
Open Compute Project
The Open Compute Project is a collaborative organization focused on sharing data center hardware designs to improve efficiency and scale.
The details
The system embeds file requests into standard NVMe commands, allowing the controller to handle searches and permissions via internal firmware. FADU plans to evolve this architecture into hostless storage servers and scalable file storage systems for the broader industry.
Timeline
The FS-SSD was unveiled in October 2026 at the Open Compute Summit.
The Tech Race
This development represents a significant evolution beyond standard NVMe block storage architectures which have dominated data center design for years. By decentralizing file management, FADU positions itself against traditional drive manufacturers in the race to optimize hardware for AI-heavy workloads.
The shift toward direct-file storage promises future improvements in processing speed and power efficiency for enterprise data centers and cloud services. Developers may eventually see reduced latency in file-heavy AI workflows as this hardware reaches widespread adoption.
The takeaway
This breakthrough signals a shift toward smarter, decentralized storage controllers that can handle complex data requests independently. Companies looking to minimize power costs and latency in AI training may prioritize these hostless storage architectures in future infrastructure deployments.
Further reading
Learn more about the latest innovations in high-performance computing hardware in our Semiconductors section.
Source note: This article includes information reported by The Korea Times.
Live Poll
Do you believe new AI storage technologies will effectively reduce national energy consumption and costs?
