{
 "id": "2.08",
 "slug": "enterprise-ssds-and-controllers",
 "url": "/parts/memory-and-storage/enterprise-ssds-and-controllers",
 "name": "Enterprise SSDs & controllers",
 "qualifiers": null,
 "layer": {
  "id": 2,
  "name": "Memory & storage"
 },
 "parent_id": null,
 "children": [
  "2.08a",
  "2.08b"
 ],
 "summary": "Solid-state drives built for data center duty cycles, and the controller chips that manage them — the fast local storage tier that feeds checkpoints, staged datasets, and cached weights to accelerators.",
 "description_md": "An enterprise SSD packages NAND (2.07) with a controller into a drive that plugs into a compute or storage tray, typically reached over the PCIe/CXL fabric in 2.06; a dense form factor sized for AI compute trays has become common as GPU-direct storage architectures pull SSD content up. AI training clusters need read and write throughput and endurance profiles distinct from general enterprise workloads, since many accelerators can be reading from or writing to storage simultaneously during checkpointing or data loading. For an investor, the drive side of this market is largely captive to the NAND makers themselves given the vertical integration described above, while the controller side has a narrower set of independent designers competing for design wins against in-house captive controllers at the largest NAND makers — and each new host interface generation is its own qualification cycle.",
 "ai_delta": "GPU-direct storage architectures and dense compute-tray form factors have pulled SSD content per AI server well above what general enterprise storage designs carry.",
 "bottleneck_status": null,
 "scale_tiers": [
  "edge",
  "enterprise",
  "hyperscale_building",
  "ai_campus",
  "gw_class"
 ],
 "density_modes": [
  "any"
 ],
 "scale_notes": "Drive count and capacity per node scale with local storage design rather than facility size; present from a single edge server up through the largest clusters.",
 "flow": [
  "compute",
  "data"
 ],
 "companies": []
}