Memory Prices

From Leadership in Tech · subscribed via aiste.ulozaite@gmail.com · original ↗ · unsubscribe

Memory Prices

DAM {“@context”:”https://schema.org”,”@type”:”WebPage”,”description”:”Historical and current prices for DRAM, HBM, and NAND flash.”,”headline”:”Memory Prices”,”url”:”http://localhost:4000/memory-prices.html”}

Stanford University

Historical and current prices for DRAM, HBM, and NAND flash.

Memory Prices

.mp-intro { margin: 0 0 1.2em; } .mp-card { border: 1px solid #e3e3e3; border-radius: 10px; padding: 1em 1.2em 0.4em; margin: 1.4em 0; box-shadow: 0 1px 4px rgba(0,0,0,.05); background:#fff; } .mp-card h2 { margin-top: .2em; } .mp-sub { color:#5a5a5a; font-size:.92em; margin:.2em 0 1em; } .mp-chart { width: 100%; height: 520px; } .mp-chart.short { height: 420px; } .mp-toolbar { display:flex; flex-wrap:wrap; gap:.6em; align-items:center; margin:.2em 0 1.2em; } .mp-btn { display:inline-block; background:#8c1515; color:#fff !important; text-decoration:none; padding:.5em .95em; border-radius:6px; font-size:.9em; font-weight:600; } .mp-btn.alt { background:#175e54; } .mp-updated { color:#5a5a5a; font-size:.88em; margin-left:auto; } .mp-note { background:#fbf6e9; border-left:4px solid #d4a017; padding:.7em 1em; border-radius:4px; font-size:.9em; margin:1em 0; } details.mp-method { margin:1.4em 0; } details.mp-method summary { cursor:pointer; font-weight:600; font-size:1.05em; } .mp-method table { width:100%; border-collapse:collapse; margin:.8em 0; font-size:.9em; } .mp-method th, .mp-method td { border:1px solid #e3e3e3; padding:.45em .6em; text-align:left; vertical-align:top; } .mp-method th { background:#f6f6f6; } .mp-err { color:#8c1515; font-weight:600; } .mp-wip { font-size:.5em; font-weight:700; letter-spacing:.03em; color:#9a6d00; background:#fbf6e9; border:1px solid #e6d8a8; border-radius:5px; padding:.2em .55em; margin-left:.55em; vertical-align:middle; text-transform:uppercase; } .mp-disclaimer { font-size:.9em; color:#5a5a5a; font-style:italic; margin:.2em 0 1em; } .mp-banner { border:1px solid #d9b38c; background:#fdf3e7; border-left:6px solid #8c1515; border-radius:6px; padding:.85em 1.15em; margin:0 0 1.2em; font-size:.96em; color:#4a3417; line-height:1.45; } .mp-banner strong { color:#8c1515; } .mp-footnote { font-size:.82em; color:#7a7a7a; margin:1.6em 0 .3em; } .mp-method h3 { margin:1.1em 0 .3em; font-size:1.02em; } .mp-method ul { margin:.3em 0 .8em 1.1em; } .mp-method li { margin:.25em 0; } .mp-toggle { display:inline-flex; margin:.1em 0 .9em; border:1px solid #c9c9c9; border-radius:6px; overflow:hidden; } .mp-toggle button { border:0; background:#f4f4f4; color:#333; padding:.38em .85em; font-size:.85em; cursor:pointer; } .mp-toggle button + button { border-left:1px solid #c9c9c9; } .mp-toggle button.active { background:#8c1515; color:#fff; }

Historic and current memory and storage prices, collected in the spirit of John C. McCallum’s classic memory-price dataset — interactive, with the raw data downloadable. Hover for details, click the legend to toggle series, drag or use the slider to zoom, and use the camera icon to export an image.

⬇ Download CSV Loading…

Price per gigabyte over time

Historical lowest $/GB on a log scale — one line per memory type: DRAM, NAND flash, and HBM. Toggle nominal vs inflation-adjusted dollars (constant 2024 $, US CPI-U).

Nominal USDReal USD (2024$)

DRAM price by generation

The DRAM line above, broken out by generation across the full history — Pre-DDR (SDRAM/core), DDR, DDR2, DDR3, DDR4, DDR5. (Generation is inferred from product descriptions, so older points are approximate.)

Nominal USDReal USD (2024$)

Accelerator cost breakdown

Modeled estimates from Epoch AI: quarterly accelerator cost across the four largest AI-accelerator designers — Nvidia, AMD, Google (TPU) and Amazon (Trainium) — stacked by component (HBM, logic die, packaging/CoWoS, auxiliary), a 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 makers on confidential contracts — there is no public spot market — so these are sparse industry-analyst estimates (TrendForce / SemiAnalysis), not transaction prices. HBM4 is projected (launches Q3 2026). $/TBps is cost per unit of memory bandwidth (stack price ÷ per-stack bandwidth).

$/GB$/TBps

Methodology note. $/GB is the cheapest listed retail price in nominal USD — not contract, average, inflation-adjusted, or a confirmed sale price. DRAM history is the McCallum dataset (extended from mid-2024 by Keepa Amazon prices); NAND is Keepa’s cheapest consumer-NVMe price from 2016 (approximate anchors before); HBM figures are modeled estimates. Sources are listed below and in the downloadable dataset; please check before citing.

Methodology, sources and caveats

Sources and method

Caveats

$/GB defaults to the cheapest retail price in nominal USD (not contract or average; retail lags contract). Use the Real USD toggle for inflation-adjusted values — constant 2024 dollars via US CPI-U annual averages (BLS).
The cheapest listing often tracks an end-of-life generation being cleared out, not the leading edge — the per-generation chart shows this.
These are cheapest listed prices over time (via Keepa), not confirmed sales. For the SSD data, obvious posting errors are removed — any month a drive is listed 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 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). The McCallum backbone and HBM estimates are fixed. The downloadable CSV lists every point with its source.

About

Compiled and maintained by David Shim, Stanford DAM project. Questions or corrections: hsshim@stanford.edu.

’’)) return GEN_PATTERNS[j][0]; return ‘older’; } // Auto-rescale a (log) y-axis to the data visible in the current x-window, so // zooming into recent years isn’t a flat line pinned to the bottom. Reusable // across charts; each keeps its own re-entrancy lock. var toMs = function (v) { return typeof v === ‘number’ ? v : new Date(v).getTime(); }; function attachYAutoscale(gd) { var lock = false; function rescale() { if (lock) return; var xr = gd._fullLayout.xaxis.range, x0 = toMs(xr[0]), x1 = toMs(xr[1]); var lo = Infinity, hi = -Infinity; gd.data.forEach(function (tr) { if (tr.visible === ‘legendonly’) return; for (var i = 0; i < tr.x.length; i++) { var t = toMs(tr.x[i]), y = tr.y[i]; if (t >= x0 && t <= x1 && y > 0) { if (y < lo) lo = y; if (y > hi) hi = y; } } }); if (!isFinite(lo)

!isFinite(hi)) return; var pad = ((Math.log10(hi) - Math.log10(lo)) * 0.06)

0.15; lock = true; Plotly.relayout(gd, { ‘yaxis.range’: [Math.log10(lo) - pad, Math.log10(hi) + pad] }) .then(function () { lock = false; }); } gd.on(‘plotly_relayout’, function (ev) { if (Object.keys(ev).some(function (k) { return k.indexOf(‘xaxis’) === 0; })) rescale(); }); rescale(); } function showStatus(msg, isErr) { STATUS.style.display = ‘block’; STATUS.innerHTML = (isErr ? ‘’ + msg + ‘’ : msg); } fetch(‘/assets/memory-prices/prices.json’, { cache: ‘no-store’ }) .then(function (r) { if (!r.ok) throw new Error(‘HTTP ‘ + r.status); return r.json(); }) .then(render) .catch(function (e) { showStatus(‘Could not load price data (‘ + e.message + ‘). The dataset may not have been ‘ + ‘generated yet — run _scripts/collect_prices.py.’, true); }); function hover(unit) { return ‘%{fullData.name}
%{x

%Y-%m-%d}
’ + ‘%{y:,.4g} ‘ + unit + ‘
%{text}’; } function render(d) { document.getElementById(‘mp-updated’).textContent = ‘Last updated: ‘ + d.generated_at; // ── Panel A: $/GB by memory TYPE — one line each for DRAM, NAND, HBM ── // Readable $ ticks: $0.01 … $100, then $1K / $1M / $1B / $1T at each power of ten. function usdTick(v) { if (v < 1e3) return ‘$’ + (+v.toPrecision(3)); if (v < 1e6) return ‘$’ + (v / 1e3) + ‘K’; if (v < 1e9) return ‘$’ + (v / 1e6) + ‘M’; if (v < 1e12) return ‘$’ + (v / 1e9) + ‘B’; return ‘$’ + (v / 1e12) + ‘T’; } var GB_TICKV = [], GB_TICKT = []; for (var _e = -2; _e <= 12; _e++) { var _v = Math.pow(10, _e); GB_TICKV.push(_v); GB_TICKT.push(usdTick(_v)); } var logY = function (extra) { return Object.assign({ title: ‘USD per GB’, type: ‘log’, gridcolor: ‘#eee’, tickvals: GB_TICKV, ticktext: GB_TICKT, automargin: true }, extra

{}); }; // US CPI-U annual averages (BLS, 1982-84=100) -> deflate $/GB into constant 2024 dollars. var CPI = {1957:28.1,1958:28.9,1959:29.1,1960:29.6,1961:29.9,1962:30.2,1963:30.6,1964:31.0, 1965:31.5,1966:32.4,1967:33.4,1968:34.8,1969:36.7,1970:38.8,1971:40.5,1972:41.8,1973:44.4, 1974:49.3,1975:53.8,1976:56.9,1977:60.6,1978:65.2,1979:72.6,1980:82.4,1981:90.9,1982:96.5, 1983:99.6,1984:103.9,1985:107.6,1986:109.6,1987:113.6,1988:118.3,1989:124.0,1990:130.7, 1991:136.2,1992:140.3,1993:144.5,1994:148.2,1995:152.4,1996:156.9,1997:160.5,1998:163.0, 1999:166.6,2000:172.2,2001:177.1,2002:179.9,2003:184.0,2004:188.9,2005:195.3,2006:201.6, 2007:207.342,2008:215.303,2009:214.537,2010:218.056,2011:224.939,2012:229.594,2013:232.957, 2014:236.736,2015:237.017,2016:240.007,2017:245.120,2018:251.107,2019:255.657,2020:258.811, 2021:270.970,2022:292.655,2023:304.702,2024:313.689,2025:322.0,2026:327.0}; var CPI_REF = 313.689; // 2024 annual average -> constant 2024 dollars function cpiOf(x) { var y = +String(x).slice(0, 4); if (y < 1957) y = 1957; if (y > 2026) y = 2026; return CPI[y]

CPI_REF; } function deflate(pp, real) { return real ? pp.map(function (p) { return { x: p.x, y: p.y * (CPI_REF / cpiOf(p.x)), text: p.text }; }) : pp; } function xZoom(g) { return (g._fullLayout && g._fullLayout.xaxis && g._fullLayout.xaxis.range) ? g._fullLayout.xaxis.range.slice() : null; } var realLab = function (real) { return real ? ‘USD per GB (constant 2024$)’ : ‘USD per GB’; }; var realUnit = function (real) { return real ? ‘USD/GB (2024$)’ : ‘USD/GB’; }; var byDate = function (a, b) { return toMs(a.x) - toMs(b.x); }; function typePts(cat) { var out = []; d.series.filter(function (s) { return s.metric === ‘usd_per_gb’ && s.category === cat; }) .forEach(function (s) { for (var i = 0; i < s.x.length; i++) out.push({ x: s.x[i], y: s.y[i], text: s.text[i]

’’, mcc: /McCallum/i.test(s.series) }); }); return out; } // DRAM = one clean line: the McCallum historical series, extended past its end by the // cheapest recent anchor per date (generations/sources are NOT split out here). var dramAll = typePts(‘DRAM’); var mcc = dramAll.filter(function (p) { return p.mcc; }).sort(byDate); var maxMcc = mcc.length ? toMs(mcc[mcc.length - 1].x) : 0; var ext = {}; dramAll.filter(function (p) { return !p.mcc && toMs(p.x) > maxMcc; }) .forEach(function (p) { if (!ext[p.x]

p.y < ext[p.x].y) ext[p.x] = p; }); var dramLine = mcc.concat(Object.keys(ext).map(function (k) { return ext[k]; })).sort(byDate); // Colorblind-safe (Okabe-Ito) colors + distinct dash/marker per series, so the // three lines are distinguishable even without relying on color. function typeTrace(name, pp, color, dash, symbol, unit) { return { name: name, type: ‘scatter’, mode: ‘lines+markers’, x: pp.map(function (p) { return p.x; }), y: pp.map(function (p) { return p.y; }), text: pp.map(function (p) { return p.text; }), hovertemplate: hover(unit

‘USD/GB’), marker: { color: color, size: 6, symbol: symbol

‘circle’ }, line: { color: color, width: 2, dash: dash

‘solid’ } }; } var gd = document.getElementById(‘chart-gb’); var GB_TYPES = [ [‘DRAM’, dramLine, ‘#0072B2’, ‘solid’, ‘circle’], [‘HBM’, typePts(‘HBM’).sort(byDate), ‘#E69F00’, ‘dash’, ‘square’], [‘NAND flash’, typePts(‘NAND’).sort(byDate), ‘#009E73’, ‘dot’, ‘diamond’] ]; function renderGB(mode) { var real = mode === ‘real’, xr = xZoom(gd); var traces = GB_TYPES.map(function (t) { return typeTrace(t[0], deflate(t[1], real), t[2], t[3], t[4], realUnit(real)); }); var layout = { margin: { t: 10, r: 16, b: 40, l: 64 }, yaxis: logY({ title: realLab(real) }), xaxis: { type: ‘date’, rangeslider: { thickness: 0.07 }, gridcolor: ‘#eee’ }, hovermode: ‘closest’, legend: { orientation: ‘h’, y: -0.25 }, paper_bgcolor: ‘#fff’, plot_bgcolor: ‘#fff’ }; if (xr) layout.xaxis.range = xr; // keep the user’s x-zoom when toggling nominal/real Plotly.newPlot(gd, traces, layout, { responsive: true, displaylogo: false, modeBarButtonsToRemove: [‘lasso2d’, ‘select2d’] }); attachYAutoscale(gd); } renderGB(‘nominal’); // ── Per-generation DRAM points (for the breakdown chart below) ───────── var pts = []; d.series.filter(function (s) { return TO_GB[s.metric] && s.category === ‘DRAM’ && s.series.indexOf(‘cheapest’) < 0; }) .forEach(function (s) { var k = TO_GB[s.metric]; for (var i = 0; i < s.x.length; i++) { pts.push({ x: s.x[i], y: s.y[i] * k, text: s.text[i]

’’, gen: genOf(s.series, s.text[i]), src: s.source }); } }); pts.sort(function (a, b) { return toMs(a.x) - toMs(b.x); }); var present = function (g) { return pts.some(function (p) { return p.gen === g; }); }; // ── Panel B: each DDR generation as its own overlapping line ─────────── var gd3 = document.getElementById(‘chart-gb-split’); var SPLIT_GENS = GEN_ORDER.filter(present); function renderGBSplit(mode) { var real = mode === ‘real’, xr = xZoom(gd3); var traces = SPLIT_GENS.map(function (g) { var P = deflate(pts.filter(function (p) { return p.gen === g; }), real); return { name: GEN_LABEL[g]

g, type: ‘scatter’, mode: ‘lines+markers’, x: P.map(function (p) { return p.x; }), y: P.map(function (p) { return p.y; }), text: P.map(function (p) { return p.text; }), hovertemplate: hover(realUnit(real)), marker: { color: GEN_COLOR[g], size: 5 }, line: { color: GEN_COLOR[g], width: 2 } }; }); var layout = { margin: { t: 10, r: 16, b: 40, l: 64 }, yaxis: logY({ title: realLab(real) }), xaxis: { type: ‘date’, rangeslider: { thickness: 0.07 }, gridcolor: ‘#eee’ }, hovermode: ‘closest’, legend: { orientation: ‘h’, y: -0.22 }, paper_bgcolor: ‘#fff’, plot_bgcolor: ‘#fff’ }; if (xr) layout.xaxis.range = xr; Plotly.newPlot(gd3, traces, layout, { responsive: true, displaylogo: false, modeBarButtonsToRemove: [‘lasso2d’, ‘select2d’] }); attachYAutoscale(gd3); } renderGBSplit(‘nominal’); // ── Accelerator cost breakdown — relative (%) and absolute ($B) ────── // Stack order HBM (bottom, largest) -> Auxiliary (top); Tableau colors; legend HBM-first. var COMP = [ { label: ‘HBM’, color: ‘#1f77b4’, share: ‘HBM cost share’, spend: ‘HBM spend’ }, { label: ‘Packaging’, color: ‘#2ca02c’, share: ‘Packaging cost share’, spend: ‘Packaging spend’ }, { label: ‘Logic’, color: ‘#ff7f0e’, share: ‘Logic cost share’, spend: ‘Logic spend’ }, { label: ‘Auxiliary’, color: ‘#7f7f7f’, share: ‘Auxiliary cost share’, spend: ‘Auxiliary spend’ } ]; function renderBreakdown(mode) { var key = mode === ‘abs’ ? ‘spend’ : ‘share’; var hov = mode === ‘abs’ ? ‘$%{y:,.4g}B’ : ‘%{y:.1f}% of BoM’; var traces = COMP.map(function (c) { var s = d.series.find(function (x) { return x.series === c[key]; }); return s ? { name: c.label, type: ‘scatter’, mode: ‘lines’, x: s.x, y: s.y, stackgroup: ‘one’, line: { width: 0.5, color: c.color }, fillcolor: c.color, hovertemplate: ‘’ + c.label + ‘
%{x

%Y-%m-%d}
’ + hov + ‘’ } : null; }).filter(Boolean); if (!traces.length) { document.getElementById(‘chart-hbm-breakdown’).innerHTML = ‘<p class="mp-sub">No data yet.</p>’; return; } var layout; if (mode === ‘abs’) { layout = { margin: { t: 10, r: 16, b: 40, l: 60 }, yaxis: { title: ‘USD billion / quarter’, gridcolor: ‘#eee’, tickprefix: ‘$’, rangemode: ‘tozero’ } }; } else { var xx = traces[0].x; // anchor the right axis (Plotly hides an overlaying axis with no trace) traces.push({ x: [xx[0], xx[xx.length - 1]], y: [0, 100], yaxis: ‘y2’, mode: ‘lines’, line: { width: 0 }, hoverinfo: ‘skip’, showlegend: false }); layout = { margin: { t: 10, r: 52, b: 40, l: 52 }, yaxis: { title: ‘% of accelerator BoM’, gridcolor: ‘#eee’, ticksuffix: ‘%’, range: [0, 100] }, yaxis2: { overlaying: ‘y’, side: ‘right’, range: [0, 100], ticksuffix: ‘%’, showgrid: false } }; } Plotly.newPlot(‘chart-hbm-breakdown’, traces, Object.assign({ xaxis: { type: ‘date’, gridcolor: ‘#eee’ }, hovermode: ‘x unified’, legend: { orientation: ‘h’, y: -0.2, traceorder: ‘normal’ }, paper_bgcolor: ‘#fff’, plot_bgcolor: ‘#fff’ }, layout), { responsive: true, displaylogo: false, modeBarButtonsToRemove: [‘lasso2d’, ‘select2d’] }); } renderBreakdown(‘abs’); // ── HBM price by generation — toggle $/GB or $/TBps, split by generation ── var HBM_GEN = [[‘HBM4’, /HBM4/i], [‘HBM3e’, /HBM3e/i], [‘HBM3’, /HBM3(?!e)/i], [‘HBM2e’, /HBM2e/i]]; var HBM_GEN_COLOR = { HBM2e: ‘#7f7f7f’, HBM3: ‘#1f77b4’, HBM3e: ‘#ff7f0e’, HBM4: ‘#2ca02c’ }; function hbmGenOf(t) { for (var i = 0; i < HBM_GEN.length; i++) if (HBM_GEN[i][1].test(t

’’)) return HBM_GEN[i][0]; return ‘HBM’; } function renderHbmGen(mode) { var isTbps = mode === ‘tbps’; var s = d.series.find(function (x) { return x.series === (isTbps ? ‘HBM $/TBps’ : ‘HBM $/GB’); }); if (!s) { document.getElementById(‘chart-hbm-gen’).innerHTML = ‘<p class="mp-sub">No data yet.</p>’; return; } var byGen = {}; for (var i = 0; i < s.x.length; i++) { var g = hbmGenOf(s.text[i]); (byGen[g] = byGen[g]

[]).push({ x: s.x[i], y: s.y[i], t: s.text[i] }); } var hov = isTbps ? ‘$%{y:,.0f} / TBps
%{text}’ : ‘$%{y:,.4g} / GB
%{text}’; var traces = [‘HBM2e’, ‘HBM3’, ‘HBM3e’, ‘HBM4’].filter(function (g) { return byGen[g]; }) .map(function (g) { var P = byGen[g]; return { name: g, type: ‘scatter’, mode: ‘lines+markers’, x: P.map(function (p) { return p.x; }), y: P.map(function (p) { return p.y; }), text: P.map(function (p) { return p.t; }), hovertemplate: ‘%{fullData.name}
%{x

%Y-%m-%d}
’ + hov + ‘’, marker: { color: HBM_GEN_COLOR[g], size: 8 }, line: { color: HBM_GEN_COLOR[g], width: 2.5 } }; }); var yaxis = isTbps ? { title: ‘USD / TBps’, tickprefix: ‘$’, gridcolor: ‘#eee’, rangemode: ‘tozero’, automargin: true } : { title: ‘USD per GB’, tickformat: ‘$,.2~f’, gridcolor: ‘#eee’, automargin: true }; Plotly.newPlot(‘chart-hbm-gen’, traces, { margin: { t: 10, r: 16, b: 40, l: 64 }, yaxis: yaxis, xaxis: { type: ‘date’, gridcolor: ‘#eee’ }, hovermode: ‘closest’, legend: { orientation: ‘h’, y: -0.2 }, paper_bgcolor: ‘#fff’, plot_bgcolor: ‘#fff’ }, { responsive: true, displaylogo: false, modeBarButtonsToRemove: [‘lasso2d’, ‘select2d’] }); } renderHbmGen(‘gb’); // Wire the toggle button groups (breakdown: abs/share; hbmgen: gb/tbps) var toggles = document.querySelectorAll(‘.mp-toggle’); for (var ti = 0; ti < toggles.length; ti++) { (function (tg) { tg.addEventListener(‘click’, function (e) { var btn = e.target.closest && e.target.closest(‘button’); if (!btn

!tg.contains(btn)) return; var bs = tg.querySelectorAll(‘button’); for (var bi = 0; bi < bs.length; bi++) bs[bi].classList.remove(‘active’); btn.classList.add(‘active’); var chart = tg.getAttribute(‘data-chart’), mode = btn.getAttribute(‘data-mode’); if (chart === ‘breakdown’) renderBreakdown(mode); else if (chart === ‘hbmgen’) renderHbmGen(mode); else if (chart === ‘units’) renderGB(mode); else if (chart === ‘units2’) renderGBSplit(mode); }); })(toggles[ti]); } // Note if the live retail source hasn’t produced data (it should, via Keepa). if (d.sources.indexOf(‘keepa’) < 0) { showStatus(‘Note: live retail data has not been collected yet — the $/GB charts currently ‘ + ‘show the historical backbone only.’, false); } } })();

Stanford
University

This site’s design borrows heavily from the Centaur project.