AI Memory Prices Surge 500% in 12 Months, Reversing Moore's Law Trends
A recent report highlights an ongoing memory crunch in the AI industry, with memory prices surging by 500% over the past year. This dramatic price hike effectively reverses the cost-reduction trends historically driven by Moore's Law, pushing economic metrics back to 2007 levels. The soaring cost of memory directly impacts the economics of AI hardware scaling, making the training and deployment of large models significantly more expensive. As High-Bandwidth Memory (HBM) demand crowds out commodity DRAM production, the broader electronics and computing industries will also face supply constraints and rising costs. The memory crunch is exacerbated by a 3-to-1 conversion ratio between HBM and DDR5 wafer capacity, meaning HBM production heavily compresses general-purpose memory supply. Major manufacturers like SK Hynix, Samsung, and Micron are shifting capacity to HBM to meet AI demand, leading to compounded price increases across all DRAM types.
## BACKGROUND
Moore's Law historically predicted that the number of transistors on a microchip doubles roughly every two years, leading to exponential increases in computing power and reductions in cost. High-Bandwidth Memory (HBM) is a 3D-stacked DRAM technology designed to provide high data transfer rates, which is critical for modern AI accelerators but requires more complex manufacturing processes than standard memory.