Transcend Launches CXL AIC Expansion Card: Up to 256GB of Memory for Compatible Devices
2026-02-25
>Innodisk, a global leader in embedded storage and industrial memory, announced the launch of its new CXL AIC expansion card. This product is designed based on the CXL architecture and uses a standard PCIe Gen 5 x8 interface to expand the system's memory capacity by up to 256GB without occupying traditional DIMM slots. It also provides high bandwidth and extremely low latency transmission performance of up to 32GB/s.


Yiding launches CXL AIC expansion card

In addition, Yiding CXL AIC expansion card adopts the compact shape of HHHL (half height and half length), which is designed for the space constrained edge computing environment. It can support two DDR5 RDIMM memory modules of standard height or VLP (ultra short version) specification, effectively save the internal space of the chassis, help the edge server and system break through the existing hardware architecture constraints, and calmly cope with the growing memory requirements of edge AI and real-time computing scenarios.

CXL low latency and high bandwidth, empowering edge applications such as 5G communication, high-frequency trading, and micro data centers

With the rapid development of applications such as 5G communication, AI inference, and real-time data processing, edge systems are simultaneously facing multiple challenges such as transmission latency, traffic fluctuations, and limited deployment space. At the same time, the demand for memory capacity and performance in the system continues to rise, but due to limitations in motherboard design, the number of traditional DIMM slots has gradually become a bottleneck for memory expansion. In this context, PCIe, as a mature and universal interface standard in edge systems, is becoming a key breakthrough in achieving memory expansion. Yiding has been deeply cultivating the memory market for a long time. While continuously strengthening mainstream product specifications, it also actively lays out cutting-edge technologies required for the next generation of computing architecture based on a deep understanding of edge AI application scenarios, and is committed to creating innovative solutions that can meet future computing power challenges. Yiding CXL AIC expansion card is an important achievement under this strategy - integrating the CXL architecture and the "Add in Card" design, it can directly use the existing PCIe slot to achieve memory expansion, effectively break the memory limit of traditional system architecture, and provide a more flexible, efficient and scalable hardware foundation for edge computing.

Yiding CXL AIC expansion card adopts PCIe Gen5 x8 interface, which can provide the system with an additional bandwidth of up to 32GB/s, and effectively reduce the transmission delay between CPU and memory. It is especially suitable for edge computing scenarios with strict requirements for real-time and low delay, including 5G communication, high-frequency financial transaction (HFT), micro data center, intelligent medical imaging and other applications.

In addition, the expansion card supports the CXL 2.0 Type 3 (CXL. mem/CXL. io) protocol, which enables memory pooling and resource sharing, allowing the system to dynamically allocate memory resources among multiple hosts based on actual workloads, while maintaining high performance and further improving the overall utilization efficiency of computing resources.

Exquisite HHHL exterior design enhances memory configuration flexibility for edge deployment

The Yiding CXL AIC expansion card adopts a compact HHHL design, which can be easily integrated into various systems. Users can install DDR5 RDIMM memory modules of different capacities according to their actual needs, achieving highly flexible hardware configuration and supporting up to 256GB of memory expansion. For small edge systems with more limited space, the Yiding DDR5 RDIMM VLP (ultra short version) memory module can also be optionally selected, effectively saving internal space of the chassis, improving wiring freedom and overall deployment flexibility.

Zhang Weimin, General Manager of Yiding International DRAM Business Unit, said, "The introduction of CXL AIC expansion card reflects our deep layout of AI edge computing. The pursuit of edge systems is not to stack hardware on a large scale like in the cloud, but to provide elastic scaling solutions that can fit specific application scenarios, balance small size, low latency, and high energy efficiency. Different from general memory products, Yiding has long been deeply involved in the field of industrial control and actively expanding edge AI applications, building a diversified product line through forward-looking technology research and development. The CXL AIC expansion card launched this time will help customers complete system upgrades in a more cost-effective way, comprehensively addressing the computing power challenges brought by low latency applications and AI inference. "

Responding to the demand for advanced computing power from cloud to end, improving the layout of CXL products
The Yiding CXL AIC expansion card and the previously launched CXL memory module are designed for different application scenarios: the CXL memory module is mainly aimed at the cloud and large data centers, using a single high-capacity, easy to plug E3. S 2T size, which is compatible with mainstream cloud server specifications; The CXL AIC expansion card focuses on small edge servers and edge AI systems with limited space, supporting users to flexibly install DDR5 memory modules of corresponding capacity on the expansion card according to actual needs, achieving more flexible system expansion.

The Yiding CXL AIC expansion card is expected to be officially sampled in the first quarter of 2026. As the CXL ecosystem gradually matures, Yiding will continue to integrate its advantages in memory, storage, expansion modules, and software and hardware collaboration, and work together with industry partners to create a more efficient and flexible next-generation system architecture.