Stanford Historical Memory Price Dataset Shows 99.9999% Cost Decline Since 1960
Tags Infrastructure · Enterprise
Stanford University published a comprehensive dataset tracking memory prices from 1960 to 2026, illustrating the extraordinary decline in storage costs over six decades. The data shows memory costs have fallen by roughly eight orders of magnitude since 1960, from dollars per byte to fractions of a cent per gigabyte. The dataset provides a rare long-term view of one of technology's most consistent exponential trends, with implications for understanding the economics of AI training, data center design, and digital storage. The visualization has sparked widespread discussion among technologists about the sustainability of current storage cost trajectories.
Technical significance
The dataset provides empirical grounding for discussions about AI infrastructure costs. Understanding historical memory price trends helps model the future cost trajectory of training and inference, and informs decisions about data retention policies and storage architecture for AI workloads.