AI BriefingThe History of Memory Prices from 1960 to 2026
Key point
A dataset visualizing historical price trends for major memory and storage types like DRAM, NAND, and HBM, along with the cost structure of AI accelerators.
Details
An interactive dataset showing price trends for DRAM, NAND Flash, and HBM from the 1960s through 2026. It provides the lowest retail price per GB for each memory type on a log scale, allowing you to track price changes across generations (DDR, HBM, etc.).
It also includes modeled data on AI accelerator cost structures. Based on Epoch AI's estimates, it breaks down the accelerator costs of major designers including Nvidia, AMD, Google (TPU), and Amazon (Trainium) by component (HBM, logic die, packaging/CoWoS, other).
For HBM (High Bandwidth Memory), since there is no public market, the dataset provides estimated prices and cost per bandwidth ($/TBps) from HBM2e through HBM4 (expected to launch in 2026), based on analysis from TrendForce and SemiAnalysis.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Memory Prices Surge 500% Over 12 Months
Next-Generation Memory Technologies Preparing for the Post-HBM Era: The Rise of Magnonics and Vertical FeRAM
High-Bandwidth Flash, Efficient Storage for Model Weight StorageMemory has grown to nearly two-thirds of AI chip component costsMemory shortage is forcing a repricing of consumer electronics