{
  "schemaVersion": 1,
  "id": "atlas-lesson-coldplate",
  "title": "Liquid-cooling coldplate",
  "category": "learning",
  "format": "Atlas lesson",
  "description": "Captures heat at the processor and transfers it into flowing coolant.",
  "canonicalUrl": "https://semiaifoundry.com/learn/atlas/coldplate/",
  "publishedDate": null,
  "revisionDate": "2026-10-07",
  "author": null,
  "topic": "Thermal path",
  "series": null,
  "license": null,
  "rightsUrl": "https://semiaifoundry.com/ai-atlas/licenses.html",
  "version": "1.3.0",
  "status": "Published educational lesson",
  "evidenceScope": "Fluid properties, flow distribution and two-phase effects are omitted.",
  "contentCoverage": "published-lesson",
  "artifactUrl": "https://semiaifoundry.com/ai-atlas/lesson-data/coldplate-ee27baac4cdcf2f3.json",
  "interactiveUrl": "https://semiaifoundry.com/ai-atlas/#coldplate",
  "related": [],
  "supersedes": [],
  "supersededBy": [],
  "contentHash": "6736ed68a3a1cbb70da84bdfb357815f904bcb8448b1ccc46668ae7c9281ad5b",
  "contentLength": 5474,
  "jsonUrl": "https://semiaifoundry.com/agents/publications/atlas-lesson-coldplate.json",
  "markdownUrl": "https://semiaifoundry.com/agents/publications/atlas-lesson-coldplate.md",
  "sections": [
    {
      "id": "section-1",
      "title": "Liquid-cooling coldplate",
      "start": 29,
      "end": 351
    },
    {
      "id": "section-2",
      "title": "The mechanism",
      "start": 351,
      "end": 1483
    },
    {
      "id": "section-3",
      "title": "Science and assumptions",
      "start": 1483,
      "end": 1511
    },
    {
      "id": "section-4",
      "title": "Flow and temperature rise account for captured heat",
      "start": 1511,
      "end": 2530
    },
    {
      "id": "section-5",
      "title": "Engineering example",
      "start": 2530,
      "end": 2671
    },
    {
      "id": "section-6",
      "title": "Distinctions and tradeoffs",
      "start": 2671,
      "end": 3235
    },
    {
      "id": "section-7",
      "title": "Specifications and interfaces",
      "start": 3235,
      "end": 3597
    },
    {
      "id": "section-8",
      "title": "Interfaces",
      "start": 3597,
      "end": 3792
    },
    {
      "id": "section-9",
      "title": "Check your understanding",
      "start": 3792,
      "end": 3821
    },
    {
      "id": "section-10",
      "title": "Why can very small channels create a cooling tradeoff?",
      "start": 3821,
      "end": 4275
    },
    {
      "id": "section-11",
      "title": "Evidence and scope",
      "start": 4275,
      "end": 4512
    },
    {
      "id": "section-12",
      "title": "Sources",
      "start": 4512,
      "end": 4984
    },
    {
      "id": "section-13",
      "title": "Continue exploring",
      "start": 4984,
      "end": 5007
    },
    {
      "id": "section-14",
      "title": "Foundations",
      "start": 5007,
      "end": 5198
    },
    {
      "id": "section-15",
      "title": "Related lessons",
      "start": 5198,
      "end": 5474
    }
  ],
  "citations": [
    {
      "title": "Explore in Atlas",
      "url": "https://semiaifoundry.com/ai-atlas/#coldplate"
    },
    {
      "title": "CoolIT: Split-Flow microchannel coldplate technology",
      "url": "https://www.coolitsystems.com/wp-content/uploads/2024/05/Split-Flow-Technology-CoolIT-Tech-Brief.pdf"
    },
    {
      "title": "CoolIT: The coldplate in a complete cooling system",
      "url": "https://www.coolitsystems.com/coldplate-technology/"
    },
    {
      "title": "Published lesson data",
      "url": "https://semiaifoundry.com/ai-atlas/lesson-data/coldplate-ee27baac4cdcf2f3.json"
    },
    {
      "title": "Atlas releases and licensing",
      "url": "https://semiaifoundry.com/ai-atlas/licenses.html"
    },
    {
      "title": "Thermal interface material",
      "url": "https://semiaifoundry.com/learn/atlas/thermal-interface/"
    },
    {
      "title": "Lid and heat spreader",
      "url": "https://semiaifoundry.com/learn/atlas/heat-spreader/"
    },
    {
      "title": "GPU logic dies",
      "url": "https://semiaifoundry.com/learn/atlas/gpu-die/"
    },
    {
      "title": "Compute baseboard and tray",
      "url": "https://semiaifoundry.com/learn/atlas/compute-baseboard/"
    },
    {
      "title": "Coolant distribution unit and separated loops",
      "url": "https://semiaifoundry.com/learn/atlas/cdu/"
    }
  ],
  "recordHash": "c551c0860d0baed8ef02b7712a6722a71cee5578a4a38c521595a52f5d328ebc",
  "contentFormat": "text/markdown",
  "content": "Thermal path · Atlas v1.3.0\n\n# Liquid-cooling coldplate\n\nCaptures heat at the processor and transfers it into flowing coolant.\n\n[Explore in Atlas](https://semiaifoundry.com/ai-atlas/#coldplate)Centimeter-scale plate with submillimeter channel features; schematic geometry\n\nLearning objectiveRelate coolant flow and heat capacity to temperature rise.\n\n## The mechanism\n\nA coldplate places a network of channels close to the heat source. Heat conducts through its metal base and enters coolant moving across the internal surfaces. Increasing contact area with small channels can improve heat transfer, but those channels also resist flow. The plate, pump, plumbing, coolant, and allowed temperatures therefore have to be designed together. Coolant travels from the inlet through channels near hot regions and leaves through the outlet, carrying the absorbed heat. Coolant temperature describes thermal energy transport; the flowing liquid is separate from the electrical power path. The warmed coolant still carries energy that must be transferred through the rack and facility cooling chain; the plate is only the point where that journey begins.\n\n1.  Heat conducts from the package through the plate’s base into the internal channel surfaces.\n2.  Coolant flowing over those surfaces absorbs thermal energy and warms as it crosses the plate.\n3.  The outlet carries warmed coolant to the larger cooling loop, where energy is transferred onward and the fluid is conditioned for reuse.\n\n## Science and assumptions\n\n### Flow and temperature rise account for captured heat\n\n$$\n\\dot Q=\\dot m c_p(T_{\\mathrm{out}}-T_{\\mathrm{in}})\n$$\n\nTransported heat equals mass flow times specific heat times outlet-to-inlet temperature rise.\n\nSymbols and units\n\nQ̇\n\nHeat transported by the coolant\n\nW\n\nṁ\n\nCoolant mass flow rate\n\nkg/s\n\nc\\_p\n\nCoolant specific heat capacity\n\nJ/(kg·K)\n\nT\\_out\n\nCoolant outlet temperature\n\n°C\n\nT\\_in\n\nCoolant inlet temperature\n\n°C\n\nṁ is coolant mass flow in kilograms per second and cp is specific heat capacity in J/(kg·K). Multiplying by the outlet-to-inlet temperature rise gives heat carried by the stream in watts. This measures transported energy, not the die temperature: the chip-to-fluid resistance still determines how hot the silicon must become to transfer that heat.\n\nAssumptions Steady flow of a single-phase coolant with nearly constant heat capacity, negligible heat leakage, and no significant pressure-work or kinetic-energy change. Boiling and mixed-flow loops require a more general enthalpy balance.\n\n## Engineering example\n\nRemoving 1 kW with water-like $c_{p}$ at 0.05 kg/s implies a temperature rise of roughly 4.8 K before plant losses.\n\n## Distinctions and tradeoffs\n\nBlue and red coolant in the atlas encode temperature, not electrical charge. The liquid transports heat; the chip receives power through a separate electrical path.\n\nSmaller channels can improve heat transfer but increase pressure drop and pumping needs. Material compatibility, flow distribution, and reliable seals also shape the design.\n\nDense accelerators need continuous heat removal to sustain their operating limits. Liquid cooling makes higher rack compute density practical when the surrounding infrastructure supports it.\n\n## Specifications and interfaces\n\nHeat capture\n\nConduction followed by convection\n\nMetal receives chip heat; fluid transports it away.\n\nUseful balance\n\nQ = mass flow × heat capacity × ΔT\n\nSteady-state energy balance for the coolant stream.\n\nChannel design\n\nMicrochannel arrays\n\nCoolIT publishes split-flow examples; no exact GB200 channel geometry is asserted.\n\n### Interfaces\n\nPackage heat Input\n\nThermal energy through contact interface\n\nCoolant inlet Input\n\nConditioned fluid from server loop\n\nCoolant outlet Output\n\nWarmer fluid carrying captured heat\n\n## Check your understanding\n\n### Why can very small channels create a cooling tradeoff?\n\n1.  They remove the need for a pump.\n2.  They turn coolant into stored model weights.\n3.  They can improve heat transfer while increasing resistance to fluid flow.\n\nAnswer and reasoning\n\nThey can improve heat transfer while increasing resistance to fluid flow.\n\nMore surface area can help heat exchange, but greater pressure drop raises pumping requirements and complicates flow distribution.\n\n## Evidence and scope\n\nFluid properties, flow distribution and two-phase effects are omitted.\n\nVary total power and a lumped junction-to-coolant thermal resistance; local flow, spatial spreading and transients remain outside the model.\n\n## Sources\n\n-   [CoolIT: Split-Flow microchannel coldplate technology](https://www.coolitsystems.com/wp-content/uploads/2024/05/Split-Flow-Technology-CoolIT-Tech-Brief.pdf)\n-   [CoolIT: The coldplate in a complete cooling system](https://www.coolitsystems.com/coldplate-technology/)\n\n[Published lesson data](https://semiaifoundry.com/ai-atlas/lesson-data/coldplate-ee27baac4cdcf2f3.json) · [Atlas releases and licensing](https://semiaifoundry.com/ai-atlas/licenses.html)\n\n## Continue exploring\n\n### Foundations\n\n-   [Thermal interface material](https://semiaifoundry.com/learn/atlas/thermal-interface/)\n-   [Lid and heat spreader](https://semiaifoundry.com/learn/atlas/heat-spreader/)\n\n### Related lessons\n\n-   [GPU logic dies](https://semiaifoundry.com/learn/atlas/gpu-die/)\n-   [Compute baseboard and tray](https://semiaifoundry.com/learn/atlas/compute-baseboard/)\n-   [Coolant distribution unit and separated loops](https://semiaifoundry.com/learn/atlas/cdu/)"
}