3 articles · Updated 2026-10-02
Storage
Drives for local AI: what the spec sheet hides, how SSD and hard-drive fraud actually works, and the tests that catch it before your data is on the disk.
Test a new drive before you trust it
2026-10-02The exact commands to run on a brand-new SSD or hard drive: SMART baseline, capacity verification, a full-surface write and read, and how to tell a used drive from a new one.
How drives get faked
2026-10-02Fake capacity, copied labels, wiped usage logs, drives off a Chia farm: how storage fraud actually works, the channels it arrives through, and the checks that still catch it.
Choosing a drive for models and datasets
2026-10-02TLC versus QLC, DRAM versus HMB, the sustained-write cliff, used datacenter U.2 drives: which numbers about an SSD survive contact with a local AI workload.
Models are the easy part of a local AI box. The drive is where the money quietly goes wrong: a 96 GB model file has to come off the disk before the GPU can touch it, datasets are read at random, checkpoints write in bursts, and the market for storage is the most counterfeited corner of PC hardware.
Three pages, in the order they matter:
- How drives get faked — fake capacity, copied labels, wiped usage logs, and which channels it arrives through. The tells that survive.
- Test a new drive before you trust it — the exact commands to run on a brand-new drive, and how to tell a used drive from a new one.
- Choosing a drive for models and datasets — TLC versus QLC, DRAM versus HMB, the sustained-write cliff, used datacenter U.2, and which numbers actually mean something.
No affiliate links anywhere on this site, no sponsored drives, nothing to buy from us.