FLASH
SSD prices rose 5 percent in July, now being ~6.5 times higher than a year ago, while hard disk drive prices also rose 5 percent but only to a level 2.5x higher than a year ago.
The numbers come from hybrid parallel file systems array supplier VDURA, which maintains a Flash Volatility Index. It checks and compares prices for 30 TB TLC (3bits/cell) and QLC (4 bits per cell) SSDs, comparing them to 30 TB HDDs. VDURA finds that 30TB TLC enterprise SSDs are now priced at $22,600, up from $3,460 in Q3 2025. 30TB QLC SSDs have followed the same trajectory, reaching $18,080 versus $2,768 a year ago. But 30 TB HDDs cost $495 a year ago and are now priced at $1,216.
The cost multiple between 30TB TLC SSDs and 30TB HDDs stands at 18.6x, down from the 23.2x peak in Q1 2026 but still nearly triple the 7.0x multiple of a year ago.
VDURA’s Eric Salo, SVP of Product and Operations, said: “Flash pricing has stopped swinging wildly, but it has settled at a level that breaks the assumptions most all-flash architectures were sold on. A 5 percent monthly increase on top of a 6.5x year-over-year jump is not relief. It is confirmation that elevated flash pricing is structural, and every AI cloud and AI factory business plan needs to account for it.”
VDURA calculates that, using its Storage Economics Optimizer Tool, and modelling the same reference deployment tracked in prior updates: 25 PB of capacity delivering 1,000 GB/s of sustained read performance, sized to AMD and NVIDIA guidance for a cluster of roughly 2,000 GPUs:
An all-flash architecture built on 30TB TLC SSDs carries a total 3-year cost of $51.60 million
An all-flash architecture built on 30TB QLC SSDs carries a total 3-year cost of $48.42 million
A hybrid VDURA SSD+HDD built on 30 TB drives carries a total 3-year cost of $12.86 million
The mixed-fleet VDURA system has 5.78 PB of flash paired with 22.68 PB of HDD in a single namespace, delivering 1,040 GB/s sustained performance. It says the all-flash approach costs roughly 4x more, while the mixed fleet limits flash exposure to only where NVMe performance actually matters.
Salo says: “Storage architecture is no longer a technical preference. It is one of the biggest financial decisions in the AI infrastructure stack, second only to the GPUs themselves. Cost per token is the scoreboard, and storage is now one of the biggest levers on it.”
VDURA’s Flash Volatility Index and Storage Economics Optimizer Tool are available here, along with supporting technical analysis.