Both, allocated by job, and the allocation has changed. Flash repriced sharply through the NAND shortage, so an SSD is no longer close to a hard drive per terabyte. Buy the SSD for the system and active work, and buy considerably more hard drive than you used to for capacity.
For most of the last decade the gap between flash and mechanical storage narrowed steadily, and by the early 2020s the standard advice, just buy an SSD, the difference is small now, was reasonable for a lot of people.
That advice is out of date. NAND supply was reallocated toward higher-margin products through 2025 and 2026, and consumer flash prices rose sharply as a result, while mechanical drive pricing stayed comparatively stable. The ratio between the two widened rather than continuing to close.
The practical consequence is that guidance written before the shortage now steers people wrong in a specific direction: it tells them to buy less capacity than they need because the flash upgrade looks affordable, and it no longer is. Anyone repeating the line that SSDs are barely more expensive is quoting a market that has moved.
The case for an SSD was never really about cost per terabyte, and the shortage clarified that. It is about behaviour under random access, and that has not changed at all.
An operating system, applications and anything with many small reads and writes are transformed by flash in a way that is obvious to whoever is using the machine. That is worth paying a premium for, and the premium is bounded because the system volume does not need to be large.
There are two other properties worth money in specific situations. No moving parts means a drop while running is survivable, which matters for anything carried. And no spin-up delay means immediate response on connection, which matters for a drive plugged in and out all day.
Buy an SSD large enough for the operating system, applications and the working set you touch regularly. For most people that is 500GB to 1TB, and going smaller becomes a false economy once updates and caches accumulate.
Then buy mechanical capacity generously, because it is the cheap part and because storage requirements only grow. Media, archives, backups, game libraries and project history all sit fine on a hard drive, and buying a size up usually improves cost per terabyte as well as deferring the next purchase.
The arrangement to avoid is a small SSD paired with a small hard drive, which leaves you managing space on both and getting the benefit of neither.
One recording-technology note for the mechanical half: if the drive might ever end up in an array, check the part number rather than the family name, because manufacturers change recording technology within a line without renaming the product.
| Drive | Capacity | Recording | $/TB | Price | |
|---|---|---|---|---|---|
| 695842-001 4TB SAS 6G | 4TB | CMR | $14.00 | $56 | Buy |
| HUS726040ALS210 4TB SAS 12G | 4TB | CMR | $18.68 | $75 | Buy |
| 695842-001 4TB SAS 6G LFF | 4TB | CMR | $20.56 | $82 | Buy |
| 012GYY 4TB SAS 6G Renewed | 4TB | CMR | $21.25 | $85 | Buy |
| DRMYH Compellent 4TB NL SAS Renewed | 4TB | CMR | $22.50 | $90 | Buy |
Absolutely, for the right job. The case for flash was always about behaviour under random access rather than price per terabyte, and that has not changed. What changed is that the price gap widened, so buying flash for bulk capacity is now a considerably worse deal than it was.
Several times more per terabyte, and the exact multiple moves with the market, which is why our live tables compute it rather than stating a fixed figure. The direction is the durable point: the gap widened through the shortage rather than continuing to close.
For capacity, yes. For the system volume, no. The allocation that works is enough SSD for the operating system and active work, then generous mechanical capacity for everything else. Buying one large SSD and no hard drive is expensive; buying one hard drive and no SSD is slow.
We do not forecast, and anyone giving you a date is guessing. What we can say is that the current ratio reflects supply allocation rather than a permanent change in manufacturing economics, and allocation decisions do reverse. Buy for the situation now rather than for a predicted one.