The next generation of low-power DRAM, LPDDR6, is entering the market at a pace faster than initially anticipated. Driven by a sharp surge in demand for high-performance, high-efficiency memory from the server and artificial intelligence sectors, the commercialization process for this new standard is accelerating.
Numerous leading semiconductor design firms are currently evaluating strategies to incorporate both LPDDR5X and LPDDR6 IP (intellectual property) simultaneously. Major mobile application processor designers, including Samsung Electronics and Qualcomm, also plan to support LPDDR6 starting with their next-generation products.
The core driver behind this shift is the explosive growth of AI applications. Both smartphones equipped with on-device AI and AI data centers that require continuous processing of massive datasets are demanding higher performance from DRAM. Global technology giants like Nvidia are actively advancing procurement of LPDDR products.
According to industry sources cited by Zdnet, more than half of high-performance semiconductor design companies are considering the parallel adoption of LPDDR5X and LPDDR6 IP. This demand is materializing more rapidly than expected, particularly in the design of chips utilizing 4-nanometer and more advanced process technologies.
**AI Applications Drive Surging Demand** The primary factor accelerating the adoption of LPDDR6 is AI. In smartphones, the traditional primary market for LPDDR, the integration of on-device AI is creating an urgent need for higher-performance memory products. Consequently, mobile AP developers like Samsung Electronics and Qualcomm plan to integrate LPDDR6 IP in their upcoming chips.
Simultaneously, the rise of AI data centers, which need to process vast amounts of data continuously, is causing a sharp increase in demand for high-performance LPDDR in the server sector. Global tech leaders such as Nvidia are actively working to secure the supply of these products.
**1.5x Performance Boost, Commercialization Possible in Second Half** LPDDR6 offers a significant performance leap over its predecessor. This generation supports bandwidths ranging from 10.6 Gbps to 14.4 Gbps, compared to the LPDDR5X range of 8.5 Gbps to a maximum of 10.7 Gbps, representing an approximate 1.5-fold increase in performance.
Although the foundational standard has been established, full-scale commercialization of LPDDR6 still requires preparation time. Supporting infrastructure, including the physical layer (PHY), controllers, and interface IP, is not yet fully ready. A formal commercial launch is expected no sooner than the second half of this year.
Currently, LPDDR6 can achieve practical performance of around 12.8 Gbps, with potential to reach 14.4 Gbps by next year. To this end, major domestic and international companies are intensively advancing their IP development efforts.
**Advanced Process Chips to Lead Adoption** LPDDR is a DRAM standard that prioritizes power efficiency more than general-purpose DDR. The 7th generation LPDDR5X is already commercially available, while LPDDR6 was finalized as the next-generation standard in July of last year.
Despite the supporting ecosystem not being fully mature, a significant number of AI and high-performance computing semiconductor designers are already progressing with plans to incorporate LPDDR6. Their strategy involves initially adopting LPDDR5X, with a performance upgrade to LPDDR6 planned once it achieves mass production scale.
According to industry sources cited by Zdnet, over half of high-performance semiconductor design firms are currently considering a parallel adoption strategy for LPDDR5X and LPDDR6 IP. The emergence of this demand has been faster than anticipated, especially among designers working on chips using 4-nanometer and more advanced process nodes.
Comments