We do not predict prices. This page shows where hard drives prices actually are, what our own daily snapshots have done, the supply-side facts we have cited, and a directional outlook that carries the date it applies to and the list of things that would prove it wrong. As of 2026-08-14 the floor is $14.00/TB across 344 in-stock listings.
Live figures updated · refreshed every 4-5 hours · outlook reviewed monthly
Daily average price per TB across all tracked SATA enterprise HDDs, daily floor TB, tracked since 2026-03-23. Live data from the DatacenterDisk price tracker — updates every 2 hours. Window reflects the actual history in our database, not a modeled projection.
Every row says where its numbers came from. Reported rows are figures published by others, cited and dated - we did not measure them. Measured rows are our own series, computed from our daily snapshots. Mixing the two without saying which is which is how a page ends up citing itself.
| What | Then | Now | Provenance |
|---|---|---|---|
| Our tracked hard drives floor, per TB | $11.33 (2026-03-23) | $12.50 (2026-08-14) | Measured by us |
As of 2026-08-14, 57 of 248 scored hard drives sit above their own recent typical level, 153 sit at the cheap end of their own record, and 38 are priced where they usually are. A further 16 are still building history. That distribution is the answer to “has waiting been working” - it is a count of badge states, not a price comparison.
In-stock only, cheapest per terabyte first.
| Product | Capacity | Condition | Price | Per TB | |
|---|---|---|---|---|---|
| HP 695842-001 4TB SAS 6G | 4TB | New | $55.99 | $14.00 | Buy |
| HP ST31000640SS 1TB SAS | 1TB | Refurb | $14.99 | $14.99 | Buy |
| HGST HUS724030ALS640 3TB SAS 6G | 3TB | Refurb | $45.39 | $15.13 | Buy |
| Seagate 3TB Enterprise Capacity SAS 6G | 3TB | Refurb | $53.95 | $17.98 | Buy |
| HGST HUS726040ALS210 4TB SAS 12G | 4TB | Refurb | $72.52 | $18.13 | Buy |
| Dell NWCCG 6TB SAS 6G NL Renewed | 6TB | Refurb | $110.00 | $18.33 | Buy |
| MDD 6TB SAS 12G 7200RPM | 6TB | Refurb | $109.99 | $18.33 | Buy |
| HP 695842-001 4TB SAS 6G LFF | 4TB | New | $78.67 | $19.67 | Buy |
Every figure in this section is read from our own catalogue at the moment you loaded the page, and it is a LEVELS observation rather than a rate. The floor is the single cheapest in-stock listing per terabyte; it can move because one cheap listing appeared or vanished, without the market having done anything. That is why percentage changes on this site come from a matched-model index that compares each listing against its own prior price, and never from comparing one day's floor to another's.
As of 2026-08-14, the floor sits at $14.00/TB and the median across 344 in-stock listings at $42.20/TB. The gap between those two is worth as much as either number on its own: a floor far below the median means the cheap end is one or two listings deep, and a floor close to it means the whole catalogue is priced tightly together.
The two condition tracks price separately and we never merge them. New stock starts at $14.00/TB and the refurbished track at $14.99/TB. A refurbished floor well below the new one is the ordinary state of this market rather than a bargain signal, and scoring one against the other would manufacture a discount that does not exist.
The picks table further down carries the same data at row level, in stock only, so a reader who decides to act can do so without leaving the page. It never shows an out-of-stock row, which is why you will not find an empty price cell anywhere in this cluster.
Our snapshots for this category begin on 2026-03-23 and run to 2026-08-14, which is 145 days of daily observations. That window is stated because a trend without its window is not checkable, and because 145 days is a real but modest history - long enough to show a level, short enough that a single unusual month shapes it.
Over the most recent 30 days of that series - 2026-07-15 to 2026-08-14 - the daily floor moved from $11.33/TB to $12.50/TB, a change of +10.3%. Read that as what the cheapest listing did, not as what the market did.
The chart is a levels series by design. A levels chart answers 'what does the cheap end cost' and it answers it honestly; it does not answer 'has the market moved', because a single listing entering or leaving changes it. Both questions are legitimate and they need different instruments, so we do not use one chart to imply the other.
Hard drive supply has been unusually tight through 2026, and the reasons are structural rather than seasonal. Lead times on enterprise nearline drives have been reported in the region of 52 weeks or more - a buyer placing an order is being quoted delivery a year out, which is a different market from one where stock is simply low.
Western Digital has been reported sold out of nearline capacity, with allocation rather than availability determining who receives drives. When a manufacturer moves to allocation, price stops being set by what buyers will pay and starts being set by who is at the front of the queue, and the retail channel is not at the front of that queue.
The capacity relief that exists is HAMR, the recording technology that lets a platter hold more without a bigger drive. It is shipping, and it raises the ceiling on what a single drive can hold. What it does not do is add manufacturing lines quickly - a new recording technology increases capacity per unit rather than units per month, and the constraint in this market is units.
Those three facts describe the same picture from different angles: demand for high-capacity nearline drives exceeds what the industry can produce, and the mechanism that clears the gap is price and queue position rather than volume.
Directionally, and stated as of the date above: the conditions that pushed drive prices up have not reversed. Lead times measured in quarters do not shorten because retail demand softens, and allocation does not end because a quarter closes. Our own floor has not been behaving like a market in retreat.
That is a statement about direction and about the absence of a reversal signal. It is not a prediction of a price, a percentage or a date, and this page will not print one. Anyone offering you a specific figure for what a terabyte will cost in six months is guessing with more confidence than the available facts support.
The more useful framing for a buyer is that the risk is asymmetric right now. In a market with long lead times and allocation, the downside of waiting is not merely that you pay the same later - it is that the specific capacity you want may not be purchasable at the moment you decide.
This is the section that makes the rest of the page worth citing. An outlook with no stated falsifiers cannot be wrong, and a claim that cannot be wrong is not information. Here is exactly what we are watching, and what each of these moving would mean:
Lead times shortening materially in reported industry figures. The 52-week number is the clearest single indicator in this category; if it comes down, the tightness is easing regardless of what any individual price does.
A sustained fall in our own daily floor - not a single cheap listing appearing for a day, but the floor holding lower across weeks. Our chart is a levels series precisely so this is visible.
In-stock breadth recovering. A market where more of the catalogue is buyable at once is a market with slack in it, and we track that count on every category page.
Any of the manufacturers moving off allocation for nearline capacity. That would be reported, and it would change the mechanism that is currently setting price.
We check these monthly and publish the result as a dated check-in whether or not the read changed. An outlook that never revises is not being tested.
If you need capacity now and the price clears your threshold, the case for buying now is stronger in this category than in most, because the constraint is supply rather than sentiment. Waiting in an allocated market risks availability as well as price.
If you are speculating on a fall, the falsifiers above are what you should be watching, and none of them has fired. That is not the same as saying prices will rise - it is saying the reasons they rose are still in place.
Either way, buy on cost per terabyte rather than on sticker price, and check the refurbished track: in this category the decommissioned-fleet supply is deep and it prices independently of the new-drive constraint.
These are the things that would change the outlook above. They are checked in every monthly edition, and the edition says so whether or not any of them moved.
Every monthly edition, newest first, archived rather than overwritten. Each one opens by stating what the previous edition said and what our data actually did - an archive that never grades itself is a blog, not a record.
The main forecast page covers the method and the hard drive picture in full · what counts as a good price · the cheapest per terabyte right now · SSD forecast · server RAM forecast · the storage crisis tracker follows the category-level picture.
We will not print a number, and you should be sceptical of anyone who does. What we can say directionally, as of the date on this page: none of the conditions that raised prices has reversed. Lead times on enterprise nearline drives have been reported at 52 weeks or more, a major manufacturer has been reported sold out of nearline capacity, and HAMR raises capacity per drive rather than drives per month. The falsifiers section above lists exactly what would change that read.
The live floor is at the top of this page and in the picks table, updated every few hours from our own catalogue. It is a levels figure - the single cheapest in-stock listing - so it can move because one listing appeared or vanished rather than because the market moved. That is why percentage changes elsewhere on this site use a matched-model index instead.
The honest answer depends on whether you need the capacity. In a market running on allocation and year-long lead times, waiting risks availability as well as price, which is a different risk from an ordinary market. If you are buying anyway, buy on cost per terabyte and consider the refurbished track. If you are purely speculating on a fall, watch the falsifiers rather than the headlines.
Demand for high-capacity nearline drives has outrun what the industry can produce. When that happens, manufacturers allocate rather than sell freely, and the retail channel sits behind the datacentre buyers in that queue. The reported 52-week-plus lead times and sold-out nearline capacity are the visible symptoms of that.
It raises the ceiling on what a single drive holds, which helps a buyer who needs a lot of capacity in few bays. It does not add production lines, and the binding constraint in this market is units per month rather than terabytes per unit. Treat it as capacity relief on a long horizon rather than a supply fix.
The live figures refresh every few hours with our price data, and the outlook is reviewed monthly - each edition is published as a dated check-in that says what moved and whether the read changed. The date beside every outlook statement on this page is the date it applies to.