Conductive-Bridge RRAM (CBRAM) is being developed to meet reliability specifications by GlobalFoundries for typical digital storage. Filamentary growth and rupture produce variability. We explore whether digital-storage-focused CBRAM can support analog current readout-based Multiply-and-Accumulate operations to enable artificial neural network applications. We explore the 2T-1R bit cell with experimental CBRAM data and GlobalFoundries’ 22FDX platform to demonstrate > 𝟐 × reduction in HRS and LRS logscale variability and > 𝟏𝟎 × higher HRS/LRS ratio. The strategy is tested for MNIST and FMNIST using system-level modeling of non-idealities. Such bit-cell design principles have general utility in exploiting variability-prone characteristics of emerging memories.