SSD Price Chart

The latest weekly median is up 2.3% from the prior successful point. The combined chain-linked median trend is 149% higher than the Aug 2025 backcast point. Demand is increasing. According to Micron's June 2026 investor presentation, AI context-memory storage is expanding the SSD market and industry data-center DRAM and NAND bit shipments in 2026 are expected to more than double their level two years earlier. SK hynix likewise reports that agentic AI and long-context processing are increasing demand for fast, high-capacity enterprise NAND storage. That demand primarily lands in enterprise and data-center SSDs rather than the consumer M.2 2280 drives in this basket. The link to consumer NVMe prices is upstream: enterprise and client drives draw on NAND output from the same manufacturers, so data-center demand can affect product mix, capacity investment, and available NAND supply. Promotions, channel inventory, and product transitions can still make retail M.2 prices move differently in any given week. SK hynix on AI-NAND and enterprise SSD demand.

The latest successful weekly point is Sep 28, 2026: a $164.71 median and $170.90 mean per nominal TB, based on 3 eligible products. These are observed US retail listings, not completed transaction prices.

It measures observed US retail listing prices for a controlled basket of new internal M.2 2280 NVMe PCIe 4.0 SSDs in 1TB and 2TB capacities. It is a finished-drive retail index, not a wholesale NAND quote or a cheapest-product list.

Start with the default median USD per TB view for the observed dollar level. Use the trend index for composition-adjusted proportional movement among overlapping products. The two can diverge when the tracked basket changes. The dashed historical line is a monthly six-drive reconstruction; the solid live line is the weekly MPC basket. The vertical marker identifies the source and cadence handoff.

Each capacity series is rebased to 100 at the August 10, 2026 live handoff. An index of 110 would mean the overlapping basket is 10% above that reference level; it would not mean an SSD costs $110.

The mean reacts to every included product and can reveal a broad early move. The median is less sensitive to one unusually high or low listing. Comparing both helps show whether movement is widespread or concentrated in a few drives.

Advanced view plots the observed median with 5-, 20-, and 200-calendar-day simple moving averages plus a 20-day volatility band. Sparse published values carry forward between observations for these calculations, so the overlays describe this series rather than forecast SSD prices.

Possible drivers include NAND supply, data-center and client demand, manufacturer product mix, controller and packaging costs, retailer inventory, and fewer promotions. The chart shows the resulting consumer listing movement; it cannot prove that one upstream factor caused a particular weekly change. Micron Q3 2026 investor presentation.

Our current view, as of Sep 28, 2026, is that a broad fall in mainstream consumer SSD listings is unlikely through the rest of 2026. Micron reported in June 2026 that industry DRAM and NAND demand continued to significantly exceed supply and expected tight conditions beyond 2027. Retail Gen4 SSDs can diverge because of promotions, channel inventory, and product changes, so this is an informed opinion rather than a price promise. Micron Q3 2026 investor presentation.

Indirectly, but not because one HBM stack replaces one SSD chip. HBM is stacked DRAM, while SSDs use NAND flash. Strong HBM demand can influence memory manufacturers’ capital spending, cleanroom expansion, advanced-packaging investment, and product priorities; SSD pricing still depends on separate NAND output and the consumer retail channel. Micron Q3 2026 Form 10-Q.

They both begin with silicon wafers, but they are different memory technologies with different process flows and manufacturing capacity. HBM uses DRAM dies and advanced stacking; SSDs use 3D NAND plus a controller. Treat industry capital and fab construction as connected constraints, not the world’s memory output as one interchangeable wafer pool. Micron Q3 2026 Form 10-Q.

More NAND wafer starts can increase potential supply, while yield, layer count, die density, node transitions, and product mix determine how many usable storage bits result. New fabs take years to build and ramp. Retail SSD prices then add controllers, packaging, warranties, channel inventory, and seller decisions on top of NAND supply. Micron Q3 2026 investor presentation.

Yes, especially through data-center storage used for model data, retrieval, checkpoints, and inference context. According to Micron, AI context-memory storage is expanding the SSD market and 2026 industry data-center DRAM and NAND bit shipments are expected to more than double their level two years earlier. SK hynix also reports increasing demand for fast, high-capacity NAND as agentic AI and long-context processing grow. That does not mean every consumer SSD listing moves immediately or by the same amount. Micron Q3 2026 prepared remarks.

Potentially. Enterprise and consumer drives can draw on output from the same NAND manufacturers, so strong data-center demand may influence capacity expansion and product mix. Their controllers, endurance, contracts, and sales channels differ, however, and this retail chart cannot isolate an enterprise-demand effect from other causes. SK hynix on AI-NAND and enterprise SSD demand.

No. NAND is the storage medium inside an SSD. A finished drive also includes a controller, firmware, DRAM or host-memory design, circuit board, packaging, validation, warranty, distribution, and retailer margin. Contract or spot NAND prices therefore should not be presented as the consumer SSD price.

At the Sep 28, 2026 point, the tracked median full-drive listing is $172.18 for 1TB and $291.36 for 2TB. These are basket references for comparable internal Gen4 NVMe drives, not quotes for every SSD model or seller.

Divide the listed drive price by nominal gigabytes. A $150 nominal 1TB drive is $0.15 per GB. This is capacity value only; it does not measure speed, endurance, warranty, or seller quality.

Price per TB puts 1TB and 2TB drives on the same capacity basis and produces easier-to-read dollar values than price per GB. The calculation uses nominal manufacturer capacity: listed price divided by capacity in GB, multiplied by 1,000.

Often, but not automatically. Use the 1TB and 2TB controls to compare the current basket medians per TB. Promotions and product mix can temporarily make one capacity more attractive, so check the full-drive price as well as the normalized figure.

The tracked 1TB median is $172.18 at the Sep 28, 2026 point. Treat it as a reference for comparable internal PCIe 4.0 NVMe drives, then check controller performance, endurance, warranty, seller, shipping, and compatibility before calling a specific listing good value.

No. A very low listing may involve weak endurance, low stock, an obscure seller, a different interface, or a retiring product. Capacity value matters, but so do controller behavior, sustained performance, warranty, compatibility, and return terms.

Use it as a market-context reference, then verify the exact model, capacity, interface, form factor, condition, seller, shipping, coupon terms, warranty, and return policy on the live listing. The chart is not an affiliate deal ranking.

The launch basket tracks Kingston NV3, Crucial P310, and WD_BLACK SN7100 in exact 1TB and 2TB variants. Each row shows its capacity, M.2 2280 PCIe 4.0 NVMe scope, and ASIN. Products remain subject to availability and eligibility checks.

Retail observations run Monday, Wednesday, and Friday. Each product’s valid checks are collapsed into one weekly median before the capacity and combined basket statistics are published.

Historical daily listings are collapsed to product-month medians and chain-linked using products present in adjacent months. Live checks become product-week medians. Both trend segments are rebased to 100 at the visibly marked August 2026 handoff.

Listings must pass exact product, capacity, condition, availability, demand, form-factor, interface, and plausible-price rules. Material changes can be queued for product-page verification instead of being accepted simply because a provider returned a number.

A failed run is recorded as failed and does not create a zero-price point. Old data is not republished as though it were a fresh observation; the page retains its last-success date and health context.

Yes, but composition changes are deliberate. Hard eligibility failures pause immediately; ordinary search absences receive a grace period. A qualified replacement can enter without rewriting any historical point, and prior product identity remains in the audit history.

No. The index holds internal non-heatsink M.2 2280 PCIe 4.0 NVMe drives constant. SATA, portable, M.2 2230, enterprise, Gen5, console-licensed, bundle, used, and renewed products serve different markets or include different premiums.

No. HDDs use magnetic storage and have a different performance, capacity, product, and supply structure. A useful SSD-versus-HDD price chart needs a separate HDD basket and methodology rather than mixing both into this NVMe index.