Western Digital, Ingrasys Partner for Next-Gen Fabric-Attached Storage

In This Article:

Western Digital Corporation WDC recently teamed up with Ingrasys, a subsidiary of Foxconn Technology Group, to transform the way storage and networking are deployed at scale. The initiative aims to engineer a Top-of-Rack (TOR) Ethernet Bunch of Flash (EBOF) switch with embedded storage—a new frontier in fabric-attached, disaggregated infrastructure designed to meet the surging data demands of AI and high-performance workloads.

The state-of-the-art TOR EBOF switch integrates Western Digital’s RapidFlex NVMe-oF bridge technology directly into a high-performance TOR switch. This convergence eliminates the need for separate storage networks, minimizes data movement latency and offers greater flexibility and scalability. Per the agreement, the TOR switch will be manufactured by Ingrasys, leveraging its expertise in high-density, GPU-optimized systems, while WDC will lead the architecture and go-to-market efforts, with a vision to make NVMe-oF disaggregated storage mainstream among cloud service providers (CSPs) and storage OEMs.

Inside WDC’s Next-Gen TOR Switch

The next-generation TOR switch from Ingrasys is expected to hit the market in 2027, featuring top-notch design and performance. At the core of the embedded storage architecture is WDC’s RapidFlex, offering 100G Ethernet connectivity and support for NVMe/PCIe Gen6 E3.S/L SSDs. The switch leverages NVIDIA’s Spectrum-4 ASIC, ensuring best-in-class switching performance and ultra-low latency. Offering flexible 400GbE and 800GbE cabling, the switch is designed for emerging hyperscale and enterprise data center environments.

RapidFlex NVMe-oF fabric bridge is an innovative solution for fabric-attached storage, designed for modern data centers. It stands out as the only NVMe-oF bridge that relies heavily on hardware acceleration and eliminates firmware from the data path. This enables I/O read and write operations to pass through the adapter with very low latency via direct Ethernet. This unique design supports smooth, high-speed integration of NVMe SSDs into disaggregated systems, allowing storage to scale independently from computing resources.

As generative AI, real-time analytics and high-throughput data applications become central to business operations, data center operators are under constant pressure to innovate. This flagship initiative is likely to become a key catalyst in next-generation data center design, aligning with trends like edge computing, data locality, GPU-driven workloads and software-defined infrastructure.

WDC Thrives on Gen AI Tailwinds Amid Broader Macro Challenges

The company expects the growing use of generative AI to drive a refresh cycle in client and consumer devices, boosting content growth in smartphones, gaming, PCs and more. Gen AI adoption jumped from 33% in 2023 to 65% in 2024, increasing storage demand for both HDDs and Flash. High-bandwidth memory is key for AI servers, while NAND flash powers SSDs for storing text, images and video. As AI data grows, eSSD sales rise due to their speed, reliability, and efficiency, helping reshape the storage market.

However, ongoing macroeconomic volatility and uncertainty caused by the damaging impact of levying tariffs and escalating global trade conflicts remain concerns for management in the near term, as these are likely to create demand fluctuations in areas such as the enterprise segment and parts of its distribution and retail business. Despite the broader uncertainty in the environment, demand from its hyperscale customers remains strong amid a constrained supply situation.

In February 2025, Western Digital completed the separation of its HDD and Flash businesses into two independent, publicly traded companies, each with a specific focus on its respective market. With a deep understanding of memory and storage technology, the new SanDisk is ready to meet market demands. It is well-equipped to take advantage of AI opportunities while maximizing the value of its products for both consumers and businesses.