1. The article discusses the limitations of current memory technologies, particularly flash, in supporting the rapid growth of AI applications. 2. It introduces ReRAM as a promising alternative, offering benefits such as lower power consumption, higher density, and suitability for embedding in AI SoCs. 3. The article highlights a session at the 2024 Design Automation Conference where Gideon Intrater from Weebit Nano will present on the potential of ReRAM in enabling low-power AI architectures.