{
  "schemaVersion": 1,
  "id": "atlas-lesson-power-supply",
  "title": "Power supply and power shelf",
  "category": "learning",
  "format": "Atlas lesson",
  "description": "Converts facility power into the regulated DC distribution used by the compute rack.",
  "canonicalUrl": "https://semiaifoundry.com/learn/atlas/power-supply/",
  "publishedDate": null,
  "revisionDate": "2026-10-07",
  "author": null,
  "topic": "Power delivery",
  "series": null,
  "license": null,
  "rightsUrl": "https://semiaifoundry.com/ai-atlas/licenses.html",
  "version": "1.3.0",
  "status": "Published educational lesson",
  "evidenceScope": "The efficiency identity is independent of brand. Values are synthetic, and real efficiency varies with input, load and temperature; distribution, regulator and facility losses must be accounted for at their own boundaries.",
  "contentCoverage": "published-lesson",
  "artifactUrl": "https://semiaifoundry.com/ai-atlas/lesson-data/power-supply-7f2ab40bff890e83.json",
  "interactiveUrl": "https://semiaifoundry.com/ai-atlas/#power-supply",
  "related": [],
  "supersedes": [],
  "supersededBy": [],
  "contentHash": "97d3991989f3463ff0d3a61f0a4757f8326b55f957f82037d24821f7b9f71182",
  "contentLength": 5844,
  "jsonUrl": "https://semiaifoundry.com/agents/publications/atlas-lesson-power-supply.json",
  "markdownUrl": "https://semiaifoundry.com/agents/publications/atlas-lesson-power-supply.md",
  "sections": [
    {
      "id": "section-1",
      "title": "Power supply and power shelf",
      "start": 31,
      "end": 375
    },
    {
      "id": "section-2",
      "title": "The mechanism",
      "start": 375,
      "end": 1538
    },
    {
      "id": "section-3",
      "title": "Science and assumptions",
      "start": 1538,
      "end": 1566
    },
    {
      "id": "section-4",
      "title": "Conversion loss becomes a cooling load",
      "start": 1566,
      "end": 2627
    },
    {
      "id": "section-5",
      "title": "Engineering example",
      "start": 2627,
      "end": 2848
    },
    {
      "id": "section-6",
      "title": "Distinctions and tradeoffs",
      "start": 2848,
      "end": 3414
    },
    {
      "id": "section-7",
      "title": "Specifications and interfaces",
      "start": 3414,
      "end": 3747
    },
    {
      "id": "section-8",
      "title": "Interfaces",
      "start": 3747,
      "end": 3969
    },
    {
      "id": "section-9",
      "title": "Check your understanding",
      "start": 3969,
      "end": 3998
    },
    {
      "id": "section-10",
      "title": "Why can a rack have more installed PSU capacity than its normal consumption?",
      "start": 3998,
      "end": 4528
    },
    {
      "id": "section-11",
      "title": "Evidence and scope",
      "start": 4528,
      "end": 4919
    },
    {
      "id": "section-12",
      "title": "Sources",
      "start": 4919,
      "end": 5414
    },
    {
      "id": "section-13",
      "title": "Continue exploring",
      "start": 5414,
      "end": 5437
    },
    {
      "id": "section-14",
      "title": "Foundations",
      "start": 5437,
      "end": 5544
    },
    {
      "id": "section-15",
      "title": "Related lessons",
      "start": 5544,
      "end": 5844
    }
  ],
  "citations": [
    {
      "title": "Explore in Atlas",
      "url": "https://semiaifoundry.com/ai-atlas/#power-supply"
    },
    {
      "title": "NVIDIA DGX GB rack systems hardware guide",
      "url": "https://docs.nvidia.com/dgx/dgxgb200-user-guide/hardware.html"
    },
    {
      "title": "NVIDIA DGX GB200 SuperPOD reference components",
      "url": "https://docs.nvidia.com/dgx-superpod/reference-architecture-scalable-infrastructure-gb200/latest/dgx-superpod-components.html"
    },
    {
      "title": "Published lesson data",
      "url": "https://semiaifoundry.com/ai-atlas/lesson-data/power-supply-7f2ab40bff890e83.json"
    },
    {
      "title": "Atlas releases and licensing",
      "url": "https://semiaifoundry.com/ai-atlas/licenses.html"
    },
    {
      "title": "Voltage regulator module",
      "url": "https://semiaifoundry.com/learn/atlas/voltage-regulator/"
    },
    {
      "title": "Compute baseboard and tray",
      "url": "https://semiaifoundry.com/learn/atlas/compute-baseboard/"
    },
    {
      "title": "Baseboard management controller",
      "url": "https://semiaifoundry.com/learn/atlas/management-controller/"
    },
    {
      "title": "Rack power shelf and DC busbar",
      "url": "https://semiaifoundry.com/learn/atlas/power-shelf/"
    }
  ],
  "recordHash": "22754262b43ea2f208e9db65236974be0eb08ebef1f7bf018f2dcfbeb9eea26f",
  "contentFormat": "text/markdown",
  "content": "Power delivery · Atlas v1.3.0\n\n# Power supply and power shelf\n\nConverts facility power into the regulated DC distribution used by the compute rack.\n\n[Explore in Atlas](https://semiaifoundry.com/ai-atlas/#power-supply)Rack shelf scale with multiple centimeter-scale supply modules\n\nLearning objectiveAccount for input power and conversion loss at a stated supply efficiency.\n\n## The mechanism\n\nA power supply converts incoming electrical power into a form the equipment can distribute and use. Rack systems combine several supplies in shelves, sharing load and providing redundancy. A shelf’s installed nameplate capacity is not the same as the power continuously consumed by the rack, especially when redundant capacity is reserved. Efficiency also varies with load, and conversion loss becomes heat that must be removed. Facility input passes through supplies into DC distribution and then downstream regulation, with losses and capacity limits at each boundary. A failure scenario can remove one supply path while the remaining capacity carries the load, making reliability tangible. Supply counts and redundancy arrangements belong to the named rack configuration, not to a universal power-chain diagram.\n\n1.  Incoming facility power is converted into the DC supply used by the rack distribution system.\n2.  Multiple power modules share the load, with extra capacity arranged according to the required redundancy strategy.\n3.  DC power feeds downstream regulators, while telemetry and protection respond to changing demand or faults.\n\n## Science and assumptions\n\n### Conversion loss becomes a cooling load\n\n$$\n\\eta=\\frac{P_{\\mathrm{out}}}{P_{\\mathrm{in}}},\\qquad P_{\\mathrm{loss}}=P_{\\mathrm{out}}\\left(\\frac{1}{\\eta}-1\\right)\n$$\n\nEfficiency is output power divided by input power. Conversion loss is output power times reciprocal efficiency minus one.\n\nSymbols and units\n\nη\n\nConversion efficiency\n\nfraction\n\nP\\_out\n\nDelivered output power\n\nW\n\nP\\_in\n\nInput power\n\nW\n\nP\\_loss\n\nPower dissipated inside the conversion boundary\n\nW\n\nEfficiency η relates useful DC output power to electrical input. The difference is dissipated in the conversion system. At large rack loads, even a small efficiency change can alter the heat that power shelves and facility cooling must remove. Installed redundant capacity is a separate concept: it changes the available power paths, not this conservation-of-energy accounting.\n\nAssumptions Steady average power at explicitly chosen input and output boundaries. Efficiency depends on voltage, load, and operating conditions; other rack and facility losses are outside a supply-only calculation.\n\n## Engineering example\n\nSynthetic supply delivering 3 kW at 96% efficiency draws 3/0.96 = 3.125 kW and loses 0.125 kW = 125 W. That heat belongs to the supply boundary; it is additional to the 3 kW delivered downstream.\n\n## Distinctions and tradeoffs\n\nInstalled PSU capacity is not the rack’s constant power consumption. Redundancy reserves capacity, and actual load depends on equipment and operating workload.\n\nRedundancy improves availability but adds installed hardware beyond the active load requirement. Conversion efficiency, airflow, and service access affect the practical rack design.\n\nLarge AI systems are constrained by deliverable electrical power as well as chip count. Power shelves turn facility capacity into a controlled supply that accelerators can actually consume.\n\n## Specifications and interfaces\n\nDGX GB200 PSU example\n\n5.5 kW per PSU\n\nNVIDIA DGX hardware guide configuration.\n\nShelf example\n\n6 PSUs / 33 kW per shelf\n\nInstalled shelf rating, not a single processor’s power.\n\nDGX rack power\n\nApproximately 120 kW\n\nPublished rack consumption example; configuration and operating workload matter.\n\n### Interfaces\n\nFacility electrical input Input\n\nAC supply to conversion stages\n\nDC distribution Output\n\nPower to busbars and downstream regulators\n\nTelemetry/control Output\n\nStatus, power readings, and fault information\n\n## Check your understanding\n\n### Why can a rack have more installed PSU capacity than its normal consumption?\n\n1.  Some capacity is reserved for redundancy and operating margin.\n2.  The unused rating proves the rack consumes no electricity.\n3.  Power supplies permanently store spare compute instructions.\n\nAnswer and reasoning\n\nSome capacity is reserved for redundancy and operating margin.\n\nRedundant designs preserve enough capacity when a supply path fails or is serviced. The installed rating and active consumption therefore describe different things.\n\n## Evidence and scope\n\nThe efficiency identity is independent of brand. Values are synthetic, and real efficiency varies with input, load and temperature; distribution, regulator and facility losses must be accounted for at their own boundaries.\n\nNo current calculator directly models this lesson’s main mechanism. Use the library for related bounds while retaining their stated omissions.\n\n## Sources\n\n-   [NVIDIA DGX GB rack systems hardware guide](https://docs.nvidia.com/dgx/dgxgb200-user-guide/hardware.html)\n-   [NVIDIA DGX GB200 SuperPOD reference components](https://docs.nvidia.com/dgx-superpod/reference-architecture-scalable-infrastructure-gb200/latest/dgx-superpod-components.html)\n\n[Published lesson data](https://semiaifoundry.com/ai-atlas/lesson-data/power-supply-7f2ab40bff890e83.json) · [Atlas releases and licensing](https://semiaifoundry.com/ai-atlas/licenses.html)\n\n## Continue exploring\n\n### Foundations\n\n-   [Voltage regulator module](https://semiaifoundry.com/learn/atlas/voltage-regulator/)\n\n### Related lessons\n\n-   [Compute baseboard and tray](https://semiaifoundry.com/learn/atlas/compute-baseboard/)\n-   [Baseboard management controller](https://semiaifoundry.com/learn/atlas/management-controller/)\n-   [Rack power shelf and DC busbar](https://semiaifoundry.com/learn/atlas/power-shelf/)"
}