Every SSD review for the last year has included some version of the same complaint: fast in a straight-line benchmark, and then, at some unpredictable moment during ordinary use, the drive stops responding for a fraction of a second while everything else waits.
That stutter has a specific cause, and it is not the flash memory itself. Nearly every SSD sold until recently used the same handful of cheap controller chips, and those controllers manage the awkward business of writing to flash cells badly. Flash cannot simply overwrite a block the way a hard disk can. It has to erase a larger region first and then write into it, and a poorly designed controller does that housekeeping reactively, in the middle of whatever you are doing, which is exactly when you notice the pause.
Intel’s new drive uses a controller the company built itself rather than licensing one of the usual off-the-shelf parts, and based on early independent benchmarks, the difference is not subtle. Sustained random write performance, the specific workload that exposes a bad controller, is several times better than the drives that shipped through most of this year. The stutter that has defined the category’s reputation is largely gone.
This tells me the flash memory was never really the problem. The chips themselves have been good enough for a while. What was missing was a company with the resources to design proper controller silicon rather than buying whatever a smaller vendor happened to have available, and it took the company that makes the processors sitting next to these drives to bother doing it properly.
The price is still steep enough that this remains a machine for people who will pay to remove every possible source of latency, not a mainstream storage upgrade. But the technical argument against SSDs, that the category is fundamentally compromised by unpredictable pauses, just lost its strongest evidence. What is left is a pricing argument, and pricing arguments resolve themselves over time in a way that architectural flaws do not.