Memory Prices | DAM
Stanford University
Memory Prices
Historic and current memory and storage prices , collected in the spirit of<br>John C. McCallum's classic memory-price dataset — interactive, with the raw data downloadable.<br>Hover for details, click the legend to toggle series, drag or use the slider to zoom, and use the<br>camera icon to export an image.
⬇ Download CSV<br>Loading…
Price per gigabyte over time
Historical lowest $/GB on a log scale — one line per memory type:<br>DRAM , NAND flash , and HBM .
DRAM price by generation
The DRAM line above, broken out by generation across the full history —<br>Pre-DDR (SDRAM/core), DDR, DDR2, DDR3, DDR4, DDR5. (Generation is inferred from product<br>descriptions, so older points are approximate.)
Accelerator cost breakdown
Modeled estimates from Epoch AI : quarterly accelerator cost across the four<br>largest AI-accelerator designers — Nvidia, AMD, Google (TPU) and Amazon (Trainium)<br>— stacked by component (HBM, logic die, packaging/CoWoS, auxiliary), a<br>production-volume-weighted average .
Absolute ($B/quarter)Share (%)
HBM price by generation
By HBM generation (HBM2e → HBM3 → HBM3e → HBM4). HBM is sold only to accelerator<br>makers on confidential contracts — there is no public spot market — so these are<br>sparse industry-analyst estimates (TrendForce / SemiAnalysis), not transaction<br>prices. HBM4 is projected (launches Q3 2026). $/TBps is cost per unit of memory bandwidth<br>(stack price ÷ per-stack bandwidth).
$/GB$/TBps
Methodology note. $/GB is the cheapest listed retail price in nominal<br>USD — not contract, average, inflation-adjusted, or a confirmed sale price. DRAM history is<br>the McCallum dataset (extended from mid-2024 by Keepa Amazon prices); NAND is Keepa's cheapest<br>consumer-NVMe price from 2016 (approximate anchors before); HBM figures are modeled estimates.<br>Sources are listed below and in the downloadable dataset; please check before citing.
Methodology, sources and caveats
Sources and method
CategoryWhat we trackSource and methodReliability
DRAM $/GB<br>cheapest retail $/GB, overall and by generation (DDR3/DDR4/DDR5)<br>Deep history (1957–2024): the McCallum memory-price dataset<br>(jcmit.net,<br>via the Internet Archive). Mid-2024 onward: the cheapest new<br>consumer DIMM each month from Keepa (Amazon retail<br>price history), refreshed monthly.<br>Reference + live<br>NAND $/GB<br>cheapest retail SSD $/GB, 2010–present<br>2016 onward: the cheapest consumer NVMe SSD each month<br>from Keepa (Amazon retail price history), refreshed monthly;<br>SATA and enterprise/datacenter drives are excluded, and per-drive posting glitches are<br>filtered (see caveats). 2010–2016: four approximate pre-NVMe<br>anchor points (no McCallum-equivalent flash dataset exists).<br>Live + approximate<br>HBM spend and cost breakdown<br>quarterly HBM spend ($B) and each component's share (%) of the accelerator bill of<br>materials (HBM, logic, packaging, auxiliary)<br>Epoch AI<br>(CC-BY): a modeled estimate, production-volume-weighted across the four largest<br>accelerator designers (Nvidia, AMD, Google, Amazon); aggregate only, no per-company split.<br>External estimate<br>HBM $/GB by generation<br>HBM price per GB and per TB/s of bandwidth, by generation<br>Industry-analyst estimates —<br>TrendForce and<br>SemiAnalysis (HBM has no public spot market);<br>bandwidth from JEDEC/Rambus.<br>HBM4 is projected.<br>Sparse estimate
Caveats
$/GB is the cheapest retail price in nominal USD — not contract, average, or<br>inflation-adjusted, and retail lags contract pricing.
The cheapest listing often tracks an end-of-life generation being cleared out ,<br>not the leading edge — the per-generation chart shows this.
These are cheapest listed prices over time (via Keepa), not confirmed<br>sales . For the SSD data, obvious posting errors are removed — any month a drive is listed<br>more than 60% below its own typical price (e.g. a $130 SSD shown at $4) is dropped.
The DRAM line splices two sources at mid-2024 (McCallum → Keepa); a small step<br>there is expected, since Amazon's cheapest clearance can sit below McCallum's representative low.
HBM figures are modeled estimates (cost share and spend), not measured prices.
Updates
DRAM and NAND $/GB refresh monthly from Keepa; HBM updates quarterly (Epoch AI).<br>The McCallum backbone and HBM estimates are fixed. The downloadable<br>CSV lists every point with its source.
About
Compiled and maintained by David Shim, Stanford DAM project. Questions or corrections:<br>hsshim@stanford.edu.
© Stanford University.<br>Stanford, California 94305.
This site's design borrows heavily from the Centaur project.