Analysts project that LPDDR memory demand from artificial intelligence infrastructure will surge dramatically by 2027. This growth is expected to surpass the combined LPDDR requirements of the two largest smartphone manufacturers, Apple and Samsung. The shift highlights a fundamental change in memory consumption patterns as AI data centers scale rapidly.
AI data center memory needs projected to exceed smartphone market totals
NVIDIA's upcoming Vera Rubin AI platform is central to this projection. Industry estimates suggest that Vera Rubin alone will require more than 60 billion gigabytes of LPDDR memory. This figure is based on current capacity estimates and expected deployment scales rather than confirmed final specifications. The demand from this single platform is projected to exceed the total smartphone needs of Apple and Samsung combined.

Apple and Samsung are each expected to sell approximately 240 million smartphones in 2025. Their combined LPDDR requirement for these devices is estimated at 57.6 billion gigabytes. In contrast, a standard smartphone typically uses 12 gigabytes of RAM. A single Vera-CPU tray with 1.5 terabytes of RAM holds the equivalent capacity of 126 standard smartphones. This massive disparity underscores the intensity of AI hardware requirements.
AMD has confirmed that its Verano platform will support SOCAMM standards starting in 2027. NVIDIA has also tested SOCAMM 2 in its DGX Station systems and plans to adopt the standard for Vera. These developments indicate a move toward standardized memory modules in high-performance computing. The industry is preparing for a period where AI infrastructure memory needs will dominate the market.
The projection that Vera Rubin will consume LPDDR like water in 2027 comes from analyst Jukan. This view is not a confirmed final specification but a prediction based on current trends. NVIDIA has not officially confirmed the exact launch window or final memory specifications for Vera Rubin.



Discussion
0 comments
Log in to join the thread with a thoughtful take, question, or correction.