Back to Blog
Education

RAM Is a Commodity: Reading DDR4 and DDR5 by Price Per Gigabyte

Memory is the most commodity-like product in computing -- nearly interchangeable, priced by the gigabyte, and swinging on a brutal boom-bust cycle. Here is how to read the DDR market with one number, and why kit configuration quietly wrecks your cost per gigabyte.

Mark | | 11 min read
DDR4DDR5RAMMemoryDRAMPrice Per GBCommodity PricingSemiconductorsBuying Guide

Of everything inside a computer, memory is the closest thing to a pure commodity. A gigabyte of DDR5 from one major manufacturer is very nearly interchangeable with a gigabyte from another. The chips come off a handful of fabrication lines, get binned by speed, and get sold by the gigabyte into a market that behaves less like consumer electronics and more like pork bellies or crude oil — violent boom-bust cycles, brutal oversupply gluts followed by savage shortages, and pricing that can swing by a factor of three within a single cycle.

That commodity nature is exactly why one number governs the entire market: dollars per gigabyte.

$/GB = kit price / total capacity in GB

If you learn to read RAM through this single ratio — and to see the two traps that quietly distort it — you will buy memory correctly through any point in its notoriously cyclical price history.


Why DRAM is a true commodity

Most computer components have durable differentiation. A CPU has an architecture and an instruction set. A GPU has a memory subsystem and a driver stack. A hard drive has recording technology and firmware. These differences let manufacturers defend margin.

DRAM has almost none of that. Its differentiation collapses to a few axes:

  • Generation — DDR4 versus DDR5, which is dictated by your motherboard and CPU, not a preference. They are not interchangeable; the notch is even physically different.
  • Speed — the data rate (e.g., DDR5-6000), a modest performance lever for most workloads.
  • Latency — the timing numbers (CL), a second-order refinement.
  • Capacity — how many gigabytes, which is what you are actually buying.

Beyond these, one manufacturer’s DRAM is functionally another’s. There is no proprietary moat at the chip level the way there is with a CPU core or a GPU architecture. This is why DRAM pricing tracks raw supply and demand more tightly than almost any other computing product — it is a global commodity with a spot-like market, and its price is set by fab capacity and inventory, not by branding. Because your platform dictates the generation, we split the boards the same way: DDR5 ranked by $/GB for current builds and DDR4 by $/GB for the still-enormous installed base.

The consequence for buyers: the brand on the heatspreader matters far less than the price per gigabyte inside it. Chase the unit price, respect the generation your platform requires, and treat everything else as a tiebreaker.


The DRAM cycle: why memory prices whipsaw

Memory has the most punishing boom-bust cycle in computing, and understanding it is the difference between buying at the top and buying at the bottom.

The cycle works like this. Building a DRAM fabrication plant costs billions and takes years. So supply cannot respond quickly to demand. When demand rises, prices spike, because supply is fixed in the short run. High prices induce manufacturers to build capacity — but that capacity arrives years later, often all at once, and often after demand has cooled. The market floods, prices collapse below cost, manufacturers cut production and delay expansions, supply tightens again, and the next shortage begins.

This is a classic capacity cycle, the same dynamic that governs oil rigs, shipping vessels, and semiconductor fabs generally. The lag between price signal and supply response guarantees overshoot in both directions. DRAM buyers do not face a smoothly rising or falling price — they face a sawtooth, and where you buy on that sawtooth can double or halve your cost per gigabyte for identical hardware.

Layer a structural demand shock on top of the ordinary cycle — an AI buildout consuming memory (and especially high-bandwidth memory) at unprecedented scale — and the peaks get higher and the shortages get sharper. When fabs prioritize the highest-margin memory for datacenter buyers, ordinary desktop DDR can get squeezed on both supply and price even though it is a different product, because it competes for the same fab capacity.

The practical implication: memory is the one component where timing genuinely rivals selection. The same kit can be a great buy and a terrible buy months apart. A price-per-gigabyte alert is worth more here than for almost any other component — and watching where DDR5 $/GB sits against its recent range is how you tell a cyclical peak from a floor before you commit.


Trap one: kit configuration wrecks your cost per gigabyte

Here is where naive $/GB math goes wrong. Memory is sold in kits — one, two, or four modules bundled together — and the configuration matters enormously for both value and future flexibility.

Two kits can offer the same total capacity at the same price and be very different purchases:

KitConfigTotalPrice$/GBSlots used
Kit A2 x 16GB32GB$110$3.44/GB2 of 4
Kit B4 x 8GB32GB$115$3.59/GB4 of 4
Kit C1 x 32GB32GB$105$3.28/GB1 of 4

By raw $/GB, Kit C looks best. But the configuration carries consequences the unit price alone does not show:

  • Kit C (1x32) is cheapest per gigabyte and leaves three slots open for future expansion — but a single module runs in single-channel mode, which can meaningfully reduce memory bandwidth on many platforms.
  • Kit A (2x16) runs in dual-channel for full bandwidth and leaves two slots open for later expansion. Usually the sweet spot.
  • Kit B (4x8) runs dual-channel but fills every slot, so any future upgrade means throwing out modules you already bought. It also stresses the memory controller more, which can limit achievable speeds.

So the honest read is $/GB as the primary sort, with configuration as a hard constraint filter. Compute the unit price first, then eliminate configurations that break your bandwidth (single vs. dual channel) or your upgrade path (all slots filled). The cheapest gigabyte that also preserves dual-channel operation and leaves headroom is the real winner — not necessarily the lowest raw number.


Trap two: speed and latency premiums that do not pay off

The second distortion is paying a large premium for a small performance gain. Memory is sold across a wide range of speeds and latencies, and the fastest kits command outsized prices. For most workloads, the marginal real-world benefit of the top speed bins over a solid mainstream bin is small — often in the low single digits of percent — while the price premium can be substantial.

The unit-pricing discipline exposes this cleanly. Compute $/GB across the speed range for your platform. When a faster kit costs 30% more per gigabyte for a few percent of real performance, the premium is being paid for a spec sheet, not for results. There are workloads that genuinely benefit from memory speed — certain simulations, integrated-graphics systems that share system memory, latency-sensitive tasks — but they are the exception. Buy enough speed to clear your platform’s sweet spot, then optimize the unit price.


How to buy memory correctly

  1. Fix your generation and platform limits first. DDR4 or DDR5 is dictated by your board and CPU. Note your maximum supported speed and slot count — these are hard constraints, not preferences.
  2. Rank candidates by dollars per gigabyte. Divide kit price by total capacity. Sort ascending.
  3. Filter by configuration. Prefer kits that preserve dual-channel operation and leave slots open for expansion. Eliminate all-slots-filled kits unless you are certain you will never upgrade.
  4. Do not overpay for speed. Buy to your platform’s sweet spot; treat top-bin kits as a premium that rarely pays for itself.
  5. Respect the cycle. Memory prices swing hard. If prices are clearly at a cyclical peak and your need is not urgent, set a $/GB alert and wait. If a genuine shortage threatens availability, buy what you need before it vanishes.

Mark’s Take: RAM is the market that most rewards treating hardware like the commodity it is. There is no meaningful moat in a gigabyte of DRAM, which is precisely why the discipline works so cleanly: normalize to dollars per gigabyte, respect the generation and channel constraints, ignore the heatspreader marketing, and time the cycle. The only thing that separates a smart memory buyer from a dumb one is whether they bought at the top or the bottom of a sawtooth they should have seen coming. Watch the unit price, watch the cycle, and let the two tell you when to move.


The bottom line

Memory is a commodity that happens to come in a box with RGB lighting. The lighting is not the product. The gigabyte is the product, and dollars per gigabyte is how you price it.

Read the DDR market through that one number, filter for the configuration that protects your bandwidth and your upgrade path, refuse to pay a spec-sheet premium for speed you will not feel, and respect the brutal cyclicality that makes timing matter more here than anywhere else in the machine. Do that, and you will buy memory well through every peak and trough the cycle throws at you — which, given DRAM’s history, it will keep throwing indefinitely.


MarketCrystal provides trend analysis and market commentary for informational purposes only. Nothing in this publication constitutes financial advice or purchasing recommendations. Memory prices are highly cyclical and change constantly; always verify current pricing before buying. Past pricing does not guarantee future results.


Follow the plumbing.

Enjoyed this article? Get early access to Trismegistus.

The boards are free forever. Trismegistus — our analysis engine, built on frontier AI — is coming. Join the first 100 subscribers for free founding access when it launches, plus the market read in your inbox.

MarketCrystal

An oracle for markets — the real price, in clear view.

Coverage

  • Stablecoins & Digital Money
  • CBDC & Monetary Policy
  • Blockchain Infrastructure
  • AI Infrastructure
  • Semiconductors & Memory
  • Energy Systems

About

MarketCrystal is an independent market price-analysis engine. We normalize every market to a real cost per unit -- $/TB, $/GB VRAM, $/W, $/Wh, $/gal -- across commodities, hardware, energy, and digital-money data, so the true price is always in clear view. Our AI analyst, Mark -- powered by the Trismegistus engine -- reads what those prices mean.

© 2026 MarketCrystal. All rights reserved.

Analysis only. Not financial advice.

No ads. No tracking. Rankings never sold. Why?