SSD vs Hard Drive Cost per Terabyte in 2026

What changed and why

The usual framing of this comparison is that flash is steadily catching up with spinning storage, and that the crossover is a matter of time. Through 2026 that framing stopped describing reality. Both media rose sharply, because the same AI buildout took capacity from both at once, and the ratio between them moved far less than either absolute price did.

On the flash side, TrendForce has reported contract increases every quarter through 2026 and forecasts a further 15 to 20% on NAND in Q4. On the spinning side, Seagate has described nearline exabytes as allocated into calendar 2028. Two different supply stories, one buyer, the same outcome.

Where the two media genuinely differ is in what a terabyte costs to make. A hard drive adds capacity by adding platters and areal density, which is cheap per terabyte and getting cheaper as HAMR arrives. Flash adds capacity by adding cells, which scales close to linearly in cost. That is the structural reason the gap persists at the top end and why the largest drives you can buy are still hard drives. Our HAMR roadmap covers the density side.

At the small end the comparison inverts. A 1 TB hard drive is not much cheaper than a 4 TB one, because the motor, actuator, enclosure and controller cost the same regardless. Below roughly 2 TB the fixed costs dominate and flash wins outright on price as well as on every performance measure, which is why the 2.5-inch hard drive has effectively disappeared from new machines.

Who it hits, and how hard

Bulk storage and media libraries: spinning storage still wins by a wide margin per terabyte, and the margin grows with capacity because the largest drives are hard drives. For a Plex library or a backup target, this is not close.

Working sets and application data: flash wins on every measure that is not price. If the workload is random, latency-sensitive, or would otherwise need many spindles to reach the required throughput, the per-terabyte comparison is the wrong question.

Small capacities: flash wins on price too, below roughly 2 TB, because a hard drive's fixed costs do not shrink with capacity.

NAS builders: the practical answer is usually both. Spinning drives for the array, flash for cache or for the applications that would otherwise keep the array awake. That hybrid is cheaper than all-flash and quieter and faster than all-spinning.

What to do now

1. Decide by workload first: bulk and sequential goes to spinning storage, random and latency-sensitive goes to flash.
2. Run the specific pair of prices through our cost per TB calculator, including usable capacity after formatting.
3. For an array, check the failure rate by model before optimising the last few dollars per terabyte.
4. Below 2 TB, stop comparing: buy flash.
5. For a NAS, size spinning capacity for the library and add a small flash cache rather than buying all-flash.
6. Check why SSD prices rose and why drive prices rose before concluding either medium is unusually bad value right now.

Outlook

Both sides of this comparison are forecast to stay tight. TrendForce expects NAND contract prices to keep rising into Q4 2026; Morgan Stanley models the hard drive shortage through 2028. Nothing in the published record suggests the ratio between them is about to move sharply in either direction.

We publish no forecast of our own. Everything above is someone else's published position, with the date it was published and a link to it, because a price forecast from a site that sells against those prices is worth nothing to you.