{"paper":{"title":"Designing Datacenter Power Delivery Hierarchies for the AI Era","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"AI datacenters must prioritize deployable capacity over time instead of installed megawatts.","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Alok Gautam Kumbhare, Chaojie Zhang, Fiodar Kazhamiaka, Grant Wilkins, Ricardo Bianchini","submitted_at":"2026-05-15T17:58:58Z","abstract_excerpt":"Demand for AI accelerators is rapidly increasing rack power density, with projections approaching 1MW per deployment by 2027. This poses a major challenge for datacenter power delivery designers. As power densities increase, a datacenter designed for a different target density may strand power, i.e., may be unable to use all the power that its delivery hierarchy has provisioned. Designs must remain efficient over long datacenter lifetimes and multiple hardware generations. Power utilization is particularly important as grid power capacity is a scarce resource in the AI era.\n  Designing an effi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"For AI datacenter design, the relevant planning objective is not installed megawatts, but deployable capacity over time.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The projection models for GPU, compute, and storage deployments together with operational factors grounded in Microsoft Azure production data accurately capture the joint effects of electrical topology, deployment granularity, placement policy, oversubscription, and workload mix across evolving sequences.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Develops a simulation framework showing multi-resource stranding changes deployable capacity and effective costs in AI datacenters, arguing the key metric is deployable capacity over time rather than installed megawatts.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"AI datacenters must prioritize deployable capacity over time instead of installed megawatts.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"6aeb51cccc81d974817ca7f95abd91e82832c47981d972b0ff6ca88cc786e62e"},"source":{"id":"2605.16255","kind":"arxiv","version":1},"verdict":{"id":"fb7529c1-b30e-47a5-baa3-e05dd8223f5c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T18:19:58.884570Z","strongest_claim":"For AI datacenter design, the relevant planning objective is not installed megawatts, but deployable capacity over time.","one_line_summary":"Develops a simulation framework showing multi-resource stranding changes deployable capacity and effective costs in AI datacenters, arguing the key metric is deployable capacity over time rather than installed megawatts.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The projection models for GPU, compute, and storage deployments together with operational factors grounded in Microsoft Azure production data accurately capture the joint effects of electrical topology, deployment granularity, placement policy, oversubscription, and workload mix across evolving sequences.","pith_extraction_headline":"AI datacenters must prioritize deployable capacity over time instead of installed megawatts."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.16255/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T18:31:18.684786Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T18:30:46.938279Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"shingle_duplication","ran_at":"2026-05-19T17:49:42.169787Z","status":"skipped","version":"0.1.0","findings_count":0},{"name":"citation_quote_validity","ran_at":"2026-05-19T17:49:41.782240Z","status":"skipped","version":"0.1.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T17:33:23.054242Z","status":"skipped","version":"1.0.0","findings_count":0},{"name":"external_links","ran_at":"2026-05-19T17:31:23.328292Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T17:01:55.591821Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"cited_work_retraction","ran_at":"2026-05-19T16:51:55.953487Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"de4b1d9ff13819c9112adb32ab1e6c388dc2a85e9b0f20da28a0a06ee965e75f"},"references":{"count":78,"sample":[{"doi":"","year":2024,"title":"AccuTech Communications. 2024. Best Data Center Build Out Cost: Top 5 Key Factors.https://accutechcom.com/data-center-build-out- cost/States Tier III construction typically $7–$9M per MW (illustrative","work_id":"fca4f399-a4c9-4f40-b6e7-fb14d19bfeda","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"2019.Power Redundancy Schemes for Data Centers","work_id":"e94a92ec-4691-4d68-99c3-8388122b3756","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Tor- res Arango","work_id":"c10bac52-611c-4250-b1b7-ab863c825923","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Hugo Barbalho, Patricia Kovaleski, Beibin Li, Luke Marshall, Marco Molinaro, Abhisek Pan, Eli Cortez, Matheus Leao, Harsh Patwari, Zuzu Tang, Larissa Rozales Gonçalves, David Dion, Thomas Mosci- broda","work_id":"d07d7652-2d6a-40d5-a353-7428f83de05c","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1007/978-3-","year":2019,"title":"2023.Parallel Programming for Multicore and Cluster Systems(3 ed.)","work_id":"cf4c4e77-acaa-46b4-b066-ddf045165d05","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":78,"snapshot_sha256":"cf248c5ad88137d4f01ab46be79ec8188a2b9553de71bcba630e846bd2941af2","internal_anchors":0},"formal_canon":{"evidence_count":1,"snapshot_sha256":"f360c16c667e4f0e5d6267954e9f07228a99e08eea024beddcb87286006d7e42"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}