Nvidia's upcoming Vera Rubin AI platform is projected to require more than six billion GB of Low-Power Double Data Rate memory in 2027 [1].
This shift signals a massive transition in how high-performance memory is distributed across the global tech economy. While LPDDR memory has traditionally been the domain of smartphones and mobile devices, the scale of AI inference workloads is moving this hardware into the data center at an unprecedented volume.
According to Citrini Research, the memory requirements for the Rubin platform will exceed the combined LPDDR usage of Apple and Samsung [2]. The two electronics giants have long dominated the low-power DRAM market through their global smartphone shipments, but the architectural needs of next-generation AI are creating a new center of gravity for supply chains.
The Vera Rubin architecture relies heavily on low-power DRAM to handle the intensive demands of AI inference [1]. This reliance is expected to drive the forecast of more than six billion GB [1] of memory needed by 2027 [1].
Industry analysts said that this demand will impact the global smartphone and AI-server markets as manufacturers scramble to allocate capacity [2]. The move toward LPDDR in servers suggests a priority on energy efficiency and power reduction, which are critical factors as AI clusters grow in size and heat output.
Nvidia has not commented on the specific figures provided by Citrini Research, but the projected volume represents a significant pivot in the hardware requirements for the AI era [2].
“The Rubin platform will reportedly demand more DRAM than Apple and Samsung combined.”
The projected memory demand for the Vera Rubin platform indicates that AI infrastructure is no longer just competing with other servers, but is now competing for the same specialized components used in the global mobile device market. This could lead to supply chain volatility for consumer electronics if LPDDR production is prioritized for high-margin AI data centers over smartphones.
