HDD|Hunt

SSD and RAM Prices Are Surging Too in 2026 — and Why HDDs Are Still the Cheapest Way to Store Bulk Data

By HDDHunt ·
prices analysis deals price per tb ssd comparison

Live HDD $/TB trend

Auto-updates each deploy

Median price per terabyte of new, internal 3.5″ SATA CMR drives on Amazon US, tracked across a fixed basket of 179 listings so day-to-day moves show real repricing, not a changing sample. Aug 12–Sep 9 2026.

Median now
$49.65/TB
Trend
holding flat · -0.5% over 29 days
Cheapest legit today
$23.80/TB · 4TB
$0 $15 $30 $45 $60 Aug 12Aug 26Sep 9 $50

Y-axis starts at $0 so the plateau reads honestly. The single cheapest legitimate drive today runs $23.80/TB below this typical-drive median , and the best high-capacity value sits at $25.50/TB (20TB) — see the live cheapest-$/TB ranking for today's floor by capacity.

Cite this chart

Writing about 2026 drive prices? Quote the trend and link back — CC-BY-4.0, no permission needed.

HDDHunt Live $/TB Trend (2026-08-12 to 2026-09-09): the median price per terabyte of new, internal 3.5" SATA CMR hard drives on Amazon US, measured across a fixed basket of 179 8TB+ listings, held near $50/TB — a -0.5% move over the window, with the single cheapest legitimate drive at $23.80/TB (4TB). Source: HDDHunt, https://hddhunt.com/blog/why-hdd-prices-are-rising-2026/#live-hdd-price-trend

Method: balanced-panel median $/TB of new internal 3.5″ SATA CMR HDDs on Amazon US (fixed basket of 179 8TB+ listings; $/TB floored at $22 to exclude fake-capacity USB scams). A dataset DOI is being minted on Zenodo (HDD-99); until then cite this page. The live ranking and the downloadable Price-Index carry the point-in-time floor.

If you’ve priced out a PC build, a NAS, or even just a memory upgrade in the last few months, you’ve probably noticed the shortage everyone’s been talking about with hard drives has quietly spread. It’s not just spinning disks anymore. SSDs and RAM are surging too — and for the same underlying reason.

I track storage prices for a living (it’s why this site exists), and the story in 2026 has turned into one story with three symptoms: hard drives, NAND flash, and DRAM are all being pulled up by the same rope. If you’re here trying to figure out why your SSD or memory kit costs what it does — and whether it’s worth waiting — here’s the honest picture, plus what it means for anyone who just wants to store a lot of data without overpaying.

The live chart above tracks the hard-drive side of this. The short version of the rest: the flash and memory side is climbing right alongside it, and that actually strengthens the case for buying spinning disks when what you need is cheap bulk capacity.

The AI Demand That Ate Hard Drives Is Now Eating Memory

I’ve written before about why HDD prices roughly doubled off their 2024 lows — AI datacenters buying enterprise drives by the tens of thousands, a two-company duopoly with no slack, and Western Digital and Seagate now sold out of 2026 capacity and pre-booked into 2028.

The exact same dynamic has now landed on memory. The AI buildout doesn’t just need cold storage; it needs enormous amounts of fast memory. A single AI server uses 8 to 10 times the DRAM of a traditional server and several times the NAND. So the memory makers — SK Hynix, Samsung, Micron — have pointed their wafers at the highest-margin product they make: high-bandwidth memory (HBM) for AI accelerators, which earns several times what ordinary DDR5 does per wafer.

That’s great for their margins and terrible for the rest of us, because every wafer that becomes HBM is a wafer that doesn’t become the DRAM in your next memory kit or the NAND in your next SSD. The result is the same shape as the hard-drive crunch:

  • The makers are “sold out.” SK Hynix has told investors its HBM, DRAM and NAND capacity is essentially sold out for 2026. That’s the memory-market echo of WD and Seagate being sold out of hard drives — the same sentence, a different aisle.
  • Prices jumped hard. Contract DRAM prices rose on the order of ~90% quarter-over-quarter at the sharpest point earlier in 2026, and NAND flash climbed by double digits per quarter. Samsung publicly warned of an industry-wide price surge and pushed module prices up steeply.
  • There’s no quick fix. Meaningful new fabrication capacity isn’t expected until late 2027 or 2028, and one analyst has warned the crunch could run past 2030. New fabs take years, not quarters.

The Good News (Relatively): The Pace Is Cooling

Just like hard drives, the memory surge looks to be settling onto an elevated shelf rather than accelerating forever. After the violent early-2026 jumps, quarterly increases have cooled to roughly low double digits — still rising, just less brutally. That’s the same “elevated plateau” pattern we’ve been describing on the HDD side: the spike is baked in, but the month-to-month panic has eased.

The important word is elevated. Cooling increases don’t mean falling prices — they mean prices that are staying high and creeping higher slowly. Nobody serious is forecasting a return to 2024 bargains across memory, flash, or disk while supply is contractually spoken for by AI buyers.

Why This Makes HDDs the Rational Bulk-Storage Choice

Here’s the part that matters if you came looking to store data rather than to build a gaming rig.

When SSDs get more expensive per terabyte, the cost-per-TB math swings back toward hard drives — hard. SSDs have always carried a big per-TB premium over HDDs; the 2026 flash squeeze widened that gap rather than closing it. So for the jobs where you’re paying for sheer capacity — a NAS, a backup target, a Plex or media library, cold archives — spinning disks are cheaper by a wide margin, and the gap is bigger this year than last.

That’s not an argument against SSDs. Flash still wins where it should: a boot drive, an OS and apps, a game library, a scratch or cache disk. Use SSDs where speed, size and power matter, and use HDDs where it’s just cheap terabytes you’re after. The 2026 squeeze didn’t change that division of labor — it sharpened it.

If you want to see the split in real dollars for your own situation, the SSD vs HDD cost calculator prices both ways live from today’s Amazon.com data: enter the capacity you need and your use case (bulk, boot, or cache) and it shows the total each way and the current per-TB gap. And for the pure “cheapest capacity” question, the live cheapest $/TB by capacity page shows which HDD bands lead right now — because the crunch scrambled the old “just size up to save” rule, and the best-value capacity moves around.

What To Actually Do

  1. If you need bulk capacity, buy HDDs, and buy at the value end of today’s shelf. Don’t wait for a crash — the supply is committed to AI customers across disk and flash, so there’s no near-term trigger for one. Check the best HDD deals and the live $/TB by capacity, and the buy-now-or-wait widget below reads today’s real number off the live shelf.
  2. If you need speed, buy the SSD — but size it honestly. With flash this expensive, don’t buy a giant SSD for data that never needed to be fast. Put the OS, apps and active projects on flash; push the bulk onto disk. The calculator will show you exactly where the crossover sits.
  3. Don’t panic-buy, but don’t wait for a forecast either. Prices move daily and nobody can time the bottom. The certain saving available now is buying the best-value drive today instead of the typical listing — that beats waiting on a price drop that may not come.

The Bottom Line

The 2026 storage squeeze isn’t a hard-drive problem anymore — it’s a memory-market problem that happens to touch DRAM, NAND and HDDs, because one buyer (AI) is pulling on all three at once. SSDs and RAM got more expensive; the makers are sold out; relief is years away.

For bulk storage, that lands on a simple conclusion: hard drives are still the cheapest way to store a lot of data, and the flash surge made that more true, not less. Buy the right tool for each job — flash for speed, disk for capacity — and when it’s capacity you’re paying for, let the live $/TB numbers pick the drive rather than a rule of thumb that the crunch already broke.