Rambus Inc. (NASDAQ: RMBS), a premier chip and silicon IP provider making data faster and safer, recently announced its DDR5 9600 Server RDIMM chipset designed to deliver industry-leading memory performance for next-generation data center platforms. Built around the new Rambus 6th Generation Registering Clock Driver (RCD06) with a 20% increase in bandwidth over the prior generation, this complete chipset enables RDIMMs operating at up to 9600 MT/s, unlocking a new level of bandwidth and capacity required by agentic AI, high-performance computing (HPC) and other data-intensive workloads.
The accelerating evolution of AI from model training to large-scale inference and agentic workflows is driving a structural increase in server memory requirements. These workloads are highly iterative and memory-intensive, requiring significantly higher bandwidth and capacity for orchestration and execution. For large language model inference, techniques such as key-value (KV) caching further increase memory capacity and bandwidth requirements by storing and repeatedly accessing contextual data to reduce compute overhead and accelerate token generation. At the same time, CPU platforms are increasing core counts and memory channel density, amplifying demand for faster DIMMs and higher aggregate memory bandwidth. Together, these trends are making memory performance and power delivery critical determinants of overall system throughput. The Rambus DDR5 9600 RDIMM chipset directly addresses these requirements, empowering server makers to build scalable platforms capable of supporting next-generation AI infrastructure.
“The rapidly accelerating adoption of agentic AI and AI inference workloads is driving unprecedented demand for higher memory bandwidth and capacity in the data center,” said Rami Sethi, SVP and general manager of Memory Interface Chips at Rambus. “With our DDR5 9600 RDIMM chipset featuring the new RCD06, Rambus continues to extend its leadership in high-speed memory interface solutions, enabling our customers to deliver the performance and reliability required for next-generation server platforms.”
“As data center architectures evolve to support increasingly complex workloads, memory bandwidth, latency and reliability are becoming critical system-level design considerations,” said Soo Kyoum Kim, associate VP at IDC. “Complete RDIMM chipsets from companies like Rambus address these demands by enabling higher-performance, scalable memory subsystems that can keep pace with next-generation AI and cloud infrastructures.”
The Rambus DDR5 9600 RDIMM chipset is a complete solution designed to support next-generation CPU-based server platforms. In addition to the RCD06, the chipset includes the PMIC5030 power management IC, which delivers efficient, high-current power at low voltage levels to support advanced DDR5 RDIMM configurations. The chipset also includes a Serial Presence Detect (SPD) Hub with an integrated temperature sensor, along with dedicated Temperature Sensor ICs, enabling critical module telemetry, configuration and thermal monitoring functions.
By integrating clocking, control and power management into a cohesive solution, the chipset ensures robust signal and power integrity at high data rates, reducing design complexity for memory module manufacturers and improving overall system reliability. This level of integration is particularly critical as server architectures scale to support higher processor core counts, larger memory capacity and sustained demands of continuously operating AI workloads.
More related news:
1) Sustainability and Climate Actions Dominate 2022 Data Centre Trends to Watch
2) Artificial Intelligence revolutionizing the course of Business strategies
3) Artificial Intelligence Business Conference
