First LPDDR6 5X 14.4 Gbps Memory IP Targets AI Infrastructure

Cadence has achieved a significant milestone by successfully taping out the industry’s first LPDDR6/5X memory IP system optimised to operate at 14.4Gbps, marking a substantial improvement of up to 50% compared to the previous generation of LPDDR DRAM. This breakthrough paves the way for enhanced AI infrastructure scalability to meet the increasing memory bandwidth and capacity requirements of next-generation AI LLMs, agentic AI, and other compute-intensive workloads across various sectors.

The new Cadence LPDDR6/5X memory IP system is poised to play a pivotal role in enabling leading AI, high-performance computing (HPC), and data centre customers to address the evolving demands of their applications. The IP system aligns with the JEDEC LPDDR6/5X standard and comprises an advanced PHY architecture and a high-performance controller designed to optimise power, performance, and area (PPA) while supporting both LPDDR6 and LPDDR5X DRAM protocols for enhanced flexibility.

Moreover, the memory IP system is engineered to seamlessly integrate into traditional monolithic SoCs as well as multi-die system architectures through the utilisation of the Cadence chiplet framework. This framework facilitates heterogeneous chiplet integration and has already demonstrated success with the previous LPDDR generation, which was successfully taped out in 2024.

Boyd Phelps, senior vice president and general manager of the Silicon Solutions Group at Cadence, highlighted the significance of LPDDR6 in the context of accelerating computing for AI inference. He emphasised the critical role of efficient data movement via memory interfaces in the evolving landscape of data centres, underscoring LPDDR6 as a key enabler for meeting the speed, bandwidth, power profile, and capacity requirements of AI workloads.

The LPDDR6/5X memory IP system represents a significant advancement in memory technology, drawing on Cadence’s expertise in DDR5, LPDDR5X, and GDDR7-36G product lines to deliver a high-performance, scalable, and adaptable architecture. This offering supports the LPDDR6 and LPDDR5X standards, including LPDDR5X CAMM2, and is tailored to cater to diverse markets such as AI, mobile, consumer, enterprise HPC, and cloud data centres.