Semiconductor design company FADU has announced that it participated in the 'FMS 2025' (Future of Memory and Storage) held in Santa Clara, California, from the 5th to the 7th of this month (local time) and unveiled its next-generation SSD technology strategy for the artificial intelligence (AI) era and its corporate vision for the next decade on the 6th.
On the opening day, the 5th, FADU's CEO Lee Ji-hyo and Ross Stenfort, a hardware systems engineer at Meta, served as co-keynote speakers. Lee shared the storage architecture and mid-to-long-term strategies optimized for AI data center environments under the theme 'Expanding the Boundaries of Storage: Next-Generation SSDs for AI Data Centers.'
FADU's CEO Lee Ji-hyo, marking the company's 10th anniversary this year, classified the storage market of the past decade into three generations. He explained that we have now entered the 'third generation AI SSD' era, which responds to AI workloads.
He also emphasized FADU's low-power, high-efficiency design philosophy. Instead of simply enhancing the performance of solid-state drive (SSD) controllers, he highlighted the importance of designs that consider power efficiency. As part of a technology roadmap to respond to the AI era, he explained plans for ▲ processing 100 million transactions (100M IOPS) per second and implementing support capacities of 512 terabytes (TB) ▲ development of petabyte (PB)-class SSD controllers.
FADU established a large exhibition booth on par with global corporations at this event, unveiling a wide array of next-generation storage technologies optimized for AI and cloud workloads.
This year, FADU introduced the achievements of the Gen5 SSD controller, which has become a significant factor in the market, and publicly debuted the code-named 'Sierra' (FC6161), a sixth-generation controller based on PCIe 6.0. This product boasts over twice the performance of existing Gen5 products with features such as ▲ support capacity of up to 512 terabytes (TB) ▲ sequential read/write speeds of 28.5 gigabytes (GB/s) ▲ random read performance of 6.9 million transactions (6900K IOPS) per second ▲ random write performance of 1 million transactions (1M IOPS) based on 7% over-provisioning, achieving industry-best performance with a low-power structure of under 9W.
FADU also announced the next-generation NVMe specification-based technology called 'Device Self Power Monitoring' (DSPM, TP-4199), jointly proposed by industry leaders including Meta.
This technology measures and analyzes real-time power data within the SSD through the integration of firmware (FW) and power management integrated circuits (PMIC). This enables accurate power predictions during the server design phase and improves energy efficiency during operation, helping to lower capital expenditures (CapEx) and operational costs (OpEx), thereby reducing the total cost of ownership (TCO).
Core technologies to meet the advanced demands of AI data center environments were also exhibited. These include ▲ Flexible Data Placement (FDP), which eliminates interference between users in multi-user environments and virtually separates the SSD to ensure consistent performance and lifespan ▲ Quad-Level Cell (QLC), a super high-density storage technology optimized for AI learning and inference servers ▲ PMIC technology that implements both power efficiency and system stability, showcasing FADU's capabilities as a leading provider of corporate SSD technologies.
Based on this, FADU plans to leverage it as a strategic foundation to expand new customer touchpoints aimed at solving practical customer challenges and responding to next-generation demands. CEO Lee Ji-hyo noted, 'Since our establishment in 2015, FADU has dedicated the past 10 years to technology development, striving to create the fastest and most innovative SSD controllers in the world targeting the global market, and we have validated our technological prowess with global clients,' adding, 'We will lead the storage market with SSDs that offer greater capacity, faster performance, and higher efficiency as demanded by the AI era.'