SPD is benefiting from the simultaneous increase in volume and price of DDR5 memory modules, particularly as AI servers are configured with more memory modules, driving sustained demand expansion. VPD is expected to open new opportunities in next-generation EDSFF enterprise solid-state drives. Domestic manufacturers such as GigaDevice (688123.SH) have taken the lead in deploying in the SPD and VPD fields, securing first-mover advantages and enhancing their competitiveness. The main points are as follows:

SPD Deeply Benefits from Memory Module Volume Expansion Some investors overlook the growth potential of SPD, which is poised to benefit significantly from the expansion in memory module volume. According to JEDEC standards, DDR5 memory modules require SPD (Serial Presence Detect), with each DDR4 and DDR5 server or PC memory module containing one SPD chip. SPD integrates EEPROM (Electrically Erasable Programmable Read-Only Memory) to store information about the memory module and configuration parameters for memory chips and related components. Additionally, SPD serves as an I2C/I3C bus hub, acting as the communication center between the system's main controller and components on the memory module. Furthermore, SPD includes an integrated TS (Temperature Sensor) that continuously monitors the temperature at its location.

In terms of value, SPD in DDR5 memory modules is more complex and valuable compared to DDR4. From a shipment perspective, global server shipments and the number of memory modules per server are continuously increasing, positioning SPD to benefit significantly from the expansion in memory module volume. According to Montage Technology's announcement, a general-purpose server is configured with 8 to 12 memory modules, while an AI server is configured with 16 to 24 memory modules. As the proportion of AI servers in global server shipments continues to rise, SPD is expected to benefit further from the expansion in memory module volume.

VPD Expected to Open Opportunities in Next-Generation eSSD VPD is set to benefit deeply from the sustained growth of the enterprise SSD market, with opportunities expanding in next-generation eSSD. VPD (Vital Product Data) is a non-volatile data storage chip used in devices or modules (such as eSSD and CXL memory expansion devices) to store critical product data, including device identification, configuration parameters, calibration data, and telemetry information.

Currently, enterprise SSDs are undergoing iterative upgrades. In the data center storage sector, the U.2 interface has long dominated the market. However, with the advent of the PCIe 6.0 era, the U.2 interface faces certain physical limitations. Consequently, EDSFF eSSD, which offers advantages in electrical performance, thermal efficiency, and storage density, is expected to become the primary choice in the PCIe 6.0 era. Leading manufacturers, such as Samsung, plan to launch 256TB sixth-generation E3.S ultra-high-density SSDs in 2027. During the iterative process of eSSD, VPD is expected to open new opportunities in next-generation EDSFF eSSD modules.

GigaDevice has leveraged its advanced R&D capabilities and strong customer base to collaborate with global leading storage manufacturers involved in setting industry standards and product specifications. The company has taken the lead in launching VPD chips for next-generation high-performance storage devices, supporting its new EDSFF eSSD modules and CXL memory modules.

Domestic SPD/VPD Manufacturers Expected to Strengthen Competitiveness In terms of SPD, GigaDevice collaborated with Montage Technology to develop DDR5 SPD chips, which passed reliability, stability, and compatibility tests by downstream customers in the second half of 2021. The chips have been widely adopted by major memory module manufacturers, establishing a first-mover advantage in this niche market and solidifying a leading position in the DDR5 SPD sector.

For VPD, GigaDevice has taken a global lead in deploying VPD chips for next-generation eSSD and CXL modules.

Risks include slower-than-expected AI adoption, slower-than-expected technological iteration, and slower-than-expected progress in domestic production.