NVIDIA announced at the Future Memory & Storage (FMS) conference that, to address the immense demand from AI for large-scale datasets and context windows, the company is introducing several advancements in storage technology. These include open-sourcing its cuFile application programming interface (API) and its underlying vertical storage software stack, which allows GPUs to directly read from and write to storage. In addition, NVIDIA is collaborating with over 40 storage industry leaders, including DDN, KIOXIA, and Micron, to advance the "Storage-Next" initiative, aimed at optimizing memory and storage solutions. The company also noted that the NVIDIA Vera CPU achieves 3.21 times higher throughput than x86 CPUs in a two-stage compression and encryption pipeline.