An analyst has warned that Nvidia's upcoming Rubin AI server platform will consume vastly more NAND flash memory than previous designs, effectively "soaking up" global supply due to a new way its systems are expected to make use of solid-state storage. This matters because it points to intensifying pressure on the NAND market from AI data-centre hardware, a trend that has already been squeezing supply and pushing up prices for components used in consumer PCs and gaming systems.

The claim centres on Rubin's architecture reportedly integrating SSD storage more deeply into the AI server design than earlier Nvidia platforms, driving demand for NAND chips well beyond typical enterprise usage. While specific figures were not detailed, the comparison to a sponge absorbing water underlines the scale of anticipated demand, raising concerns that PC gamers and other consumers could face tighter SSD availability and higher prices as manufacturers prioritise supply for AI infrastructure.