{
  "proofs_version": 1,
  "manifest_id": "manifest-ai-data-center-electricity-demand-throug-0cd6d2fe",
  "sentence_count": 169,
  "classes": {
    "grounded": 122,
    "attributed": 46,
    "verbatim": 1
  },
  "proofs": [
    {
      "section_id": "executive-brief",
      "sentence": "U.S. data centers consumed approximately 176 TWh in 2023, representing 4.4% of total U.S. electricity consumption, according to a 2024 report on U.S. data center energy usage.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "4.4%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "executive-brief",
      "sentence": "That growth accelerated sharply: LBNL documents a compound annual growth rate of approximately 7% from 2014 to 2018, rising to 18% between 2018 and 2023.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2014",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2018",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "7%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "18%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "executive-brief",
      "sentence": "LBNL's scenario range for 2028 runs from approximately 325 TWh to 580 TWh, representing 6.7% to 12.0% of forecasted U.S. electricity consumption.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "6.7%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "12.0%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "the scenario variations provide a range of total data center energy estimates, with the low and high end of roughly 325 and 580 TWh in 2028, as shown in Figure ES-1. Assuming an average capacity utilization rate of 50%, this annual energy use range would translate to a total power demand for data centers between 74 and 132 GW. This annual energy use also represents 6.7% to 12.0% of total U.S. elec",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "executive-brief",
      "sentence": "EPRI's February 2026 state-level pipeline analysis projects 384 to 793 TWh by 2030, having revised upward 60% from its 2024 estimate following 18 months of accelerated AI infrastructure announcements.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "60%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "EPRI's February",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "executive-brief",
      "sentence": "Goldman Sachs, in its February 2026 update, implies approximately 810 TWh for the U.S. alone by 2030.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Goldman Sachs",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "executive-brief",
      "sentence": "Globally, the IEA's base case projects consumption roughly doubling to around 945 TWh by 2030, growing at approximately 15% per year.[^src-iea-org-reports-energy-and-ai--cbc5f094]",
      "source_ids": [
        "src-iea-org-reports-energy-and-ai--cbc5f094"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-iea-org-reports-energy-and-ai--cbc5f094"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "15%",
          "found_in": [
            "src-iea-org-reports-energy-and-ai--cbc5f094"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "IEA",
          "found_in": [
            "src-iea-org-reports-energy-and-ai--cbc5f094"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Our Base Case finds that global electricity consumption for data centres is projected to double to reach around 945 TWh by 2030 in the Base Case, representing just under 3% of total global electricity consumption in 2030.",
      "quote_source": "src-iea-org-reports-energy-and-ai--cbc5f094"
    },
    {
      "section_id": "executive-brief",
      "sentence": "One synthesis of these forecasts places a defensible U.S. base case at 550 to 650 TWh, representing 11 to 13% of U.S. total, by blending LBNL's equipment method with EPRI's medium scenario and McKinsey's workload approach.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "13%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "EPRI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "executive-brief",
      "sentence": "LBNL's bottom-up model incorporates GPU power draws at 60 to 80% of rated capacity and projects power usage effectiveness (PUE) declining to approximately 1.4 through liquid cooling and hyperscale adoption.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "80%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.4",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "executive-brief",
      "sentence": "The IEA documents that accelerated servers are projected to grow at 30% annually in its base case, while conventional servers grow at 9%.[^src-iea-org-reports-energy-and-ai--cbc5f094]",
      "source_ids": [
        "src-iea-org-reports-energy-and-ai--cbc5f094"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "30%",
          "found_in": [
            "src-iea-org-reports-energy-and-ai--cbc5f094"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "9%",
          "found_in": [
            "src-iea-org-reports-energy-and-ai--cbc5f094"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "IEA",
          "found_in": [
            "src-iea-org-reports-energy-and-ai--cbc5f094"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Electricity consumption in accelerated servers, which is mainly driven by AI adoption, is projected to grow by 30% annually in the Base Case, while conventional server electricity consumption growth is slower at9% per year. Accelerated servers account for almost half of the net increase in global data centre electricity consumption",
      "quote_source": "src-iea-org-reports-energy-and-ai--cbc5f094"
    },
    {
      "section_id": "executive-brief",
      "sentence": "The evidence base suggests over 2,000 GW of generation and storage capacity currently sit in U.S. interconnection queues, representing more than the total existing U.S. generation fleet.[^src-data-centers-and-ai-energy-con-234e7e27]",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        }
      ],
      "quote": "OpenAI's GPT-4, released in 2023, required an estimated 50 gigawatt-hours (GWh) of electricity for its training run\u2014equivalent to the annual consumption of approximately 40,000 U.S. households or a small city's worth of power.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "executive-brief",
      "sentence": "LBNL's own methodology explicitly avoids overestimation caused by tracking load for projects that have not yet selected a power provider.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "executive-brief",
      "sentence": "Grid interconnection timelines now routinely extend up to four years.[^src-ropesgray-com-en-insights-view-cdedd4a2]",
      "source_ids": [
        "src-ropesgray-com-en-insights-view-cdedd4a2"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Participants indicated that U.S. data center power demand could reach 35\u201345 GW by 2030\u2014roughly double 2024 levels.",
      "quote_source": "src-ropesgray-com-en-insights-view-cdedd4a2"
    },
    {
      "section_id": "executive-brief",
      "sentence": "Electrical capacity is governed by physical grid limitations and utility planning cycles that often extend multiple years, unlike land or capital.[^src-data-center-resources-240b7648]",
      "source_ids": [
        "src-data-center-resources-240b7648"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "The data center industry is undergoing a fundamental structural transformation that is no longer driven primarily by real estate fundamentals, but by infrastructure constraints tied directly to energy systems. What once revolved around land acquisition, tax incentives, and geographic positioning has now shifted toward a far more complex equation involving power availability, transmission capacity,",
      "quote_source": "src-data-center-resources-240b7648"
    },
    {
      "section_id": "executive-brief",
      "sentence": "EPRI's pipeline method aggregates operational, under-construction, and announced projects before applying scenario-based realization rates.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "executive-brief",
      "sentence": "A benchmark from Grid Strategies suggests only approximately 65% of announced projects may materialize by 2030.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "65%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "executive-brief",
      "sentence": "The LBNL report announced by DOE noted that data center load is projected to double or triple by 2028, but that framing\u2014which emphasizes that clean energy and grid modernization can meet the growth\u2014does not grapple with the gap between queue entries and energized megawatts.[^src-doe-releases-new-report-evalua-762bc619]",
      "source_ids": [
        "src-doe-releases-new-report-evalua-762bc619"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-doe-releases-new-report-evalua-762bc619"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-doe-releases-new-report-evalua-762bc619"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "DOE",
          "found_in": [
            "src-doe-releases-new-report-evalua-762bc619"
          ],
          "via": "evidence"
        }
      ],
      "quote": "is projected to double or triple by 2028",
      "quote_source": "src-doe-releases-new-report-evalua-762bc619"
    },
    {
      "section_id": "executive-brief",
      "sentence": "The primary evidence for the LBNL scenarios rests on the 2024 LBNL report itself.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Many underlying models on which financial analytics firms build these projections are proprietary or methodologically opaque (Goldman Sachs Research 2024; Jeff Brown 2024; Bloomberg 2024). While not meaningless, historical utility demand forecasts consistently overestimate both peak and average demand (Carvallo et al. 2018).",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "executive-brief",
      "sentence": "LBNL's own framing is explicit: many financial analytics firm models are proprietary or methodologically opaque, and historical utility demand forecasts have consistently overestimated both peak and average demand.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "current-baseline",
      "sentence": "The LBNL report itself is the primary source for this figure, derived through the laboratory's bottom-up methodology.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ]
    },
    {
      "section_id": "current-baseline",
      "sentence": "That 176 TWh number is the most defensible current baseline available: according to the Congressional Research Service, it derives from LBNL's bottom-up methodology and represents the closest public-record approximation to a measured floor for U.S. data center energy consumption in 2023.[^src-congress-gov-crs-product-r4864-2112aaad]",
      "source_ids": [
        "src-congress-gov-crs-product-r4864-2112aaad"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-congress-gov-crs-product-r4864-2112aaad"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Congressional Research Service",
          "found_in": [
            "src-congress-gov-crs-product-r4864-2112aaad"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-congress-gov-crs-product-r4864-2112aaad"
          ],
          "via": "evidence"
        }
      ],
      "quote": "According to estimates produced by Lawrence Berkeley National Laboratory (LBNL), U.S. data center annual energy use in 2023 (not accounting for cryptocurrency) was approximately 176 terawatt-hours (TWh), approximately 4.4% of U.S. annual electricity consumption that year.",
      "quote_source": "src-congress-gov-crs-product-r4864-2112aaad"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-eta-publications-lbl-gov-sites-f3f409d8] That design choice is not incidental.",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-eta-publications-lbl-gov-sites-f3f409d8] An investor treating sell-side estimates as equivalent to LBNL's equipment-based accounting is not comparing two equally weighted sources.",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ]
    },
    {
      "section_id": "current-baseline",
      "sentence": "The source projects that IEA data show global data center electricity consumption could double by 2030, potentially reaching 600\u2013800 TWh annually, though the source carries no primary methodology of its own.[^src-data-centers-and-ai-energy-con-234e7e27]",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "IEA",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The International Energy Agency (IEA) projects that data center electricity consumption could double by 2030, potentially reaching 600-800 TWh annually.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "current-baseline",
      "sentence": "On global current consumption, the honest answer is that the floor is uncertain within a range of a few hundred TWh, and no single institution has produced a comprehensive bottom-up global count comparable to the detailed national-level methodologies that exist for individual countries.[^src-socomec-us-en-us-solutions-bus-f305877b]",
      "source_ids": [
        "src-socomec-us-en-us-solutions-bus-f305877b"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "According to the International Energy Agency, data centers currently consume between 2% and 3% of the world's electricity. In 2023, this infrastructure consumed approximately 176 terawatt-hours (TWh)",
      "quote_source": "src-socomec-us-en-us-solutions-bus-f305877b"
    },
    {
      "section_id": "current-baseline",
      "sentence": "That acceleration reflects the rapid growth in GPU-accelerated servers for AI applications entering the installed base.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "During the early and mid-2010s, a shift from on-premise data centers to colocation or cloud facilities helped enable efficiency improvements that allowed data center electricity use to remain nearly constant at a time when the data center industry grew significantly, with a large expansion of data center services. The efficiency strategies that allowed the industry to avoid increased energy needs ",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-ai-data-centers-and-the-u-s-el-73756c3b] Capital commitment at that scale is corroborating evidence that demand growth is real, not simply projected.",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 terawatt hours (TWh) in 2023 (or, about 4.4% of total U.S. electricity consumption) to between 325-580 TWh (6.7-12.0%) by 2028",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-congress-gov-crs-product-r4864-2112aaad] That is the only available figure from a named institutional source that attempts the split.",
      "source_ids": [
        "src-congress-gov-crs-product-r4864-2112aaad"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "According to estimates produced by Lawrence Berkeley National Laboratory (LBNL), U.S. data center annual energy use in 2023 (not accounting for cryptocurrency) was approximately 176 terawatt-hours (TWh), approximately 4.4% of U.S. annual electricity consumption that year.",
      "quote_source": "src-congress-gov-crs-product-r4864-2112aaad"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-data-centers-and-ai-energy-con-234e7e27] It is not part of the empirical floor.",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "The International Energy Agency (IEA) projects that data center electricity consumption could double by 2030, potentially reaching 600-800 TWh annually.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-congress-gov-crs-product-r4864-2112aaad] These claims are real data points about what is being asserted in the discourse.",
      "source_ids": [
        "src-congress-gov-crs-product-r4864-2112aaad"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "According to estimates produced by Lawrence Berkeley National Laboratory (LBNL), U.S. data center annual energy use in 2023 (not accounting for cryptocurrency) was approximately 176 terawatt-hours (TWh), approximately 4.4% of U.S. annual electricity consumption that year.",
      "quote_source": "src-congress-gov-crs-product-r4864-2112aaad"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-power-demand-forecasts-revised-66709781] These figures are grounded in regulatory filings and pipeline data, not forward projections alone.",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-power-demand-forecasts-revised-66709781] But the same analysis notes that data center market analysts from multiple independent firms suggest demand growth is unlikely to exceed 60 to 65 GW through 2030, implying utility forecasts overstate data center-driven load growth by about 40%.",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "40%",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-power-demand-forecasts-revised-66709781] The source acknowledges directly that institutional projections of data center electricity demand vary widely, with utility load forecasts implying data center growth of roughly 90 GW while independent market analysts place the ceiling closer to 60\u201365 GW through 2030\u2014a disagreement it describes as complicating medium-to-long-term grid planning.",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-ai-data-centers-and-the-u-s-el-73756c3b] A spread of that magnitude is not statistical noise.",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 terawatt hours (TWh) in 2023 (or, about 4.4% of total U.S. electricity consumption) to between 325-580 TWh (6.7-12.0%) by 2028",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    },
    {
      "section_id": "current-baseline",
      "sentence": "The same source corroborates this pattern, noting that its own 2022 review of 258 data center energy estimates from 46 publications between 2007 and 2021 found only two sources of reliable estimates, pointing to systematic methodological defects in the historical literature.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2022",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2007",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2021",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "My 2022 review of 258 data center energy estimates from 46 original publications between 2007 and 2021 showed only two sources of reliable estimates.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-ai-data-centers-and-the-u-s-el-73756c3b] [^src-eta-publications-lbl-gov-sites-f3f409d8] The methodological record does not disqualify modern estimates.",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b",
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 terawatt hours (TWh) in 2023 (or, about 4.4% of total U.S. electricity consumption) to between 325-580 TWh (6.7-12.0%) by 2028",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-data-centers-and-ai-energy-con-234e7e27] This is one source, not independent corroboration.",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "The International Energy Agency (IEA) projects that data center electricity consumption could double by 2030, potentially reaching 600-800 TWh annually.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "current-baseline",
      "sentence": "[^src-ai-data-centers-and-the-u-s-el-73756c3b][^src-eta-publications-lbl-gov-sites-f3f409d8] The 325 to 580 TWh range for 2028 is an LBNL scenario output.",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b",
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b",
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 terawatt hours (TWh) in 2023 (or, about 4.4% of total U.S. electricity consumption) to between 325-580 TWh (6.7-12.0%) by 2028",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The verified baseline established in the preceding section \u2014 176 TWh of U.S. data-center electricity consumption in 2023, representing 4.4% of total U.S. electricity use \u2014 is where consensus ends.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "4.4%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The LBNL bottom-up equipment model projects 325 to 580 TWh by 2028, with 13\u201327% CAGR extrapolations pointing toward 450 to 850 TWh by 2030.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "27%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "CAGR",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "EPRI's February 2026 pipeline analysis, revised upward 60% from its 2024 estimate following 18 months of accelerated AI infrastructure announcements, spans 384 to 793 TWh in 2030 across its low, medium, and high realization scenarios.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "60%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "EPRI's February",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Goldman Sachs updated its global forecast to a 220% growth scenario by 2030, implying roughly 810 TWh for the U.S. share.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "220%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Goldman Sachs",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "At the outer extreme, the World Resources Institute documents a projection reaching over 1,050 TWh, a figure that would represent about a quarter of all U.S. electricity generation in 2023.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "World Resources Institute",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ],
      "quote": "Modeled energy use projections through 2030 range from 200 TWh/year to over 1,050 TWh/year. That highest figure, published by Boston Consulting Group, would represent about a quarter of all U.S. electricity generation in 2023 (which was 4,178 TWh).",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Meanwhile, many estimates cluster between 300 and 400 TWh by 2030.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Modeled energy use projections through 2030 range from 200 TWh/year to over 1,050 TWh/year. That highest figure, published by Boston Consulting Group, would represent about a quarter of all U.S. electricity generation in 2023 (which was 4,178 TWh).",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Over three years, utility five-year peak load growth forecasts submitted to FERC rose from 24 GW in 2022 to 166 GW in 2025 \u2014 more than a six-fold increase \u2014 with data centers accounting for roughly 55% of that growth, or approximately 90 GW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2022",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2025",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "55%",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "FERC",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Grid Strategies' 2024 analysis found that U.S. electricity demand could rise by 128 GW over the next five years, driven primarily by data centers and manufacturing growth concentrated in six regions of the country.[^src-five-year-us-load-growth-forec-14069f2c]",
      "source_ids": [
        "src-five-year-us-load-growth-forec-14069f2c"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-five-year-us-load-growth-forec-14069f2c"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-five-year-us-load-growth-forec-14069f2c"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-five-year-us-load-growth-forec-14069f2c"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. electricity demand could rise 128 GW over the next five years, driven by data centers and manufacturing growth primarily in six regions of the country, according to a report published Thursday by Grid Strategies.",
      "quote_source": "src-five-year-us-load-growth-forec-14069f2c"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "**GPU and accelerator power-draw trajectory.** The LBNL bottom-up model anchors its estimates in equipment shipments, with GPU power draws assumed at 60 to 80% of rated capacity.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "80%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "LBNL's bottom-up equipment model projects 325-580 TWh by 2028 (6.7-12% US total), driven by AI server shipments where training overtakes inference (50-53% AI energy) amid GPU power ramps (60-80% rated) and PUE declines to ~1.4 via liquid cooling/hyperscale shift",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Pipeline-based forecasters, including EPRI, convert announced IT capacity into energy use through implicit load factors; their high scenario reaches 132 GW of IT capacity.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "EPRI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "IT",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Grid Strategies estimates that alternative benchmarks from market analysts \u2014 Cleanview, TD Cowen, Bloomberg NEF, McKinsey, S&P Global, Wood Mackenzie \u2014 suggest data-center demand growth is unlikely to exceed 60 to 65 GW through 2030, implying that utility forecasts of roughly 90 GW overstate demand by about 25 GW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "TD Cowen",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Bloomberg NEF",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "P Global",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Wood Mackenzie",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Cleanview",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "LBNL makes this explicit: its bottom-up method avoids overestimation by tracking installed equipment rather than load for projects that have not yet selected a power provider.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "LBNL's data shows that during the early and mid-2010s, the shift from on-premises infrastructure to cloud facilities enabled near-flat electricity consumption despite large service-volume growth.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The devsustainability.com analysis rejects this analogy directly: because major AI labs are already deploying into hyperscale facilities, \"the biggest historical efficiency lever \u2014 moving workloads out of inefficient enterprise data centers and into modern cloud infrastructure \u2014 has already been pulled.\"[^src-ai-data-center-energy-in-2026-fb073872] Future gains will have to come from better accelerators, model architectures, higher utilization, better scheduling, and inference optimization.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Future",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The major AI labs are already deploying into hyperscale facilities, so new capacity is highly efficient from the start. But that also means the biggest historical efficiency lever - moving workloads out of inefficient enterprise data centers and into modern cloud infrastructure - has already been pulled.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "That same analysis also notes that AI workload patterns are shifting from single-query inference through reasoning models toward agents that decompose problems into many model calls, tool calls, retrieval steps, and verification loops \u2014 complicating any fixed efficiency assumption.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The major AI labs are already deploying into hyperscale facilities, so new capacity is highly efficient from the start. But that also means the biggest historical efficiency lever - moving workloads out of inefficient enterprise data centers and into modern cloud infrastructure - has already been pulled.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The World Resources Institute points to an additional complication: efficiency gains may trigger a rebound effect that ultimately drives total demand higher despite lower power-per-task.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [
        {
          "kind": "entity",
          "value": "World Resources Institute",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "However, the available evidence on that claim rests on a single modeling exercise, not independent lines of evidence, and the empirical question of whether AI-enabled grid optimization will be deployed at sufficient scale and speed remains unresolved.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The major AI labs are already deploying into hyperscale facilities, so new capacity is highly efficient from the start. But that also means the biggest historical efficiency lever - moving workloads out of inefficient enterprise data centers and into modern cloud infrastructure - has already been pulled.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "**Cooling overhead and PUE evolution.** Average data-center PUE has declined from approximately 2.0 in the early 2000s to 1.5 to 1.6 for typical facilities today, with leading hyperscale operators reporting fleet-wide PUE of 1.10 to 1.12.[^src-data-centers-and-ai-energy-con-234e7e27]",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "number",
          "value": "2.0",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.5",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.6",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.10",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.12",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Average",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Average PUE has declined from approximately 2.0 in the early 2000s (meaning half of all energy went to non-IT functions) to 1.5-1.6 for typical facilities today. Leading hyperscale operators achieve PUE ratios of 1.1-1.3, with Google reporting fleet-wide annual PUE of 1.10 and Microsoft achieving 1.12.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Cooling can account for 40% of a data center's energy usage, making it a prime target.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "40%",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Cooling, in particular, can account for 40% of a data center's energy usage, making it a prime target for greater efficiency.",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "LBNL's scenarios assume PUE declining to approximately 1.4 through liquid cooling and hyperscale facility shifts.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "number",
          "value": "1.4",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "McKinsey's workload-driven forecast assumes PUE improvement to 1.1.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "number",
          "value": "1.1",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "**Capacity-addition completion rates.** The Luminix synthesis reports that only roughly 65% of announced data-center projects may materialize by 2030 per Grid Strategies benchmarks.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "65%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Luminix",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "EPRI's high scenario applies a 30% early-planning realization rate on top of under-construction and advanced-planning capacity; its low scenario is materially more conservative.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "30%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The World Resources Institute documents the mechanism behind non-completion: utilities are receiving speculative and duplicate interconnection requests, including early-phase projects unlikely to be built and multiple requests for the same facility, generating phantom load that distorts forecasts.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [
        {
          "kind": "entity",
          "value": "World Resources Institute",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Grid Strategies observes that the power industry does not have a clear understanding of how much demand will actually come from data centers.[^src-five-year-us-load-growth-forec-14069f2c]",
      "source_ids": [
        "src-five-year-us-load-growth-forec-14069f2c"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-five-year-us-load-growth-forec-14069f2c"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. electricity demand could rise 128 GW over the next five years, driven by data centers and manufacturing growth primarily in six regions of the country, according to a report published Thursday by Grid Strategies.",
      "quote_source": "src-five-year-us-load-growth-forec-14069f2c"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The Luminix synthesis, which positions itself as an investor-facing analysis and should be read in that promotional framing, proposes a defensible base case of 550 to 650 TWh by 2030, blending EPRI's medium scenario, McKinsey's workload-driven approach, and LBNL's trend-adjusted high, under assumptions of hyperscaler dominance at 85% of servers, PUE of 1.2 to 1.4, and roughly even training-to-inference split.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "85%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.2",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.4",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Luminix",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "EPRI",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Defensible base case: 550\u2013650 TWh (11\u201313% of ~5,000 TWh total US generation). This blends EPRI's medium scenario (596 TWh), LBNL's trend-adjusted high, and Goldman's US-share estimate, assuming hyperscaler dominance (85% of servers), PUE of 1.2\u20131.4, and a roughly even training/inference split.",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Its sensitivity analysis suggests potential variance of plus 20% if inference exceeds training, minus 15% if GPU efficiency and PUE below 1.2 are achieved, and plus-or-minus 10% from project realization risk.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "20%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "15%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "10%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.2",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The LBNL report establishes 176 TWh for 2023.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Other sources report the global baseline at roughly 300 to 400 TWh for the same period.[^src-data-centers-and-ai-energy-con-234e7e27]",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "The International Energy Agency (IEA) projects that data center electricity consumption could double by 2030, potentially reaching 600-800 TWh annually.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The devsustainability.com analysis examined 258 data-center energy estimates published between 2007 and 2021 and found that only two sources provided reliable estimates.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2007",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2021",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "My 2022 review of 258 data center energy estimates from 46 original publications between 2007 and 2021 showed only two sources of reliable estimates.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "Many historical estimates employed weak extrapolation, taking energy cost per query, multiplying by user growth, and calling it a forecast.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "LBNL explicitly distinguishes its bottom-up equipment methodology from the proprietary or methodologically opaque models underlying many financial analytics projections, and notes that historical utility demand forecasts have consistently overestimated both peak and average demand.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The World Resources Institute similarly cautions that data-center electricity demand was overestimated in the 2000s, when projections of power shortages after explosive 2000-to-2009 growth proved wrong as efficiency improvements held consumption flat through 2018.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2000",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2009",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2018",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "World Resources Institute",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ],
      "quote": "During the 2000s, the rise of Internet services caused explosive growth in development, with data center electricity use rising 90% between 2000 and 2005 and 24% between 2005 and 2009. However, between 2010 to 2018 global data center electricity use was basically flat, as more efficient technologies were introduced even as computing demand soared across the world. In hindsight, electricity demand ",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "forecast-landscape",
      "sentence": "The ratio of speculative to committed capacity \u2014 the phantom-load problem \u2014 is the mechanism behind the 25 GW gap Grid Strategies identifies between utility forecasts and market-analyst benchmarks.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The interconnection queue figure most frequently cited to anchor the scale of AI-driven power demand comes from Lawrence Berkeley National Laboratory: over 2,000 GW of generation and storage capacity currently sits in interconnection queues across the United States.[^src-data-center-grid-limitations-t-30b0465e]",
      "source_ids": [
        "src-data-center-grid-limitations-t-30b0465e"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "Lawrence Berkeley National Laboratory",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "United States",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        }
      ],
      "quote": "interconnection wait times have more than doubled over the past fifteen years, with projects now spending an average of five years in queue before reaching commercial operation",
      "quote_source": "src-data-center-grid-limitations-t-30b0465e"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Grid Strategies' analysis suggests that utility forecasts overstate data center demand by roughly 25 GW relative to third-party market benchmarks, and historical interconnection tracking finds that only 13 percent of projects requesting interconnection between 2000 and 2019 had reached commercial operation by end of 2024, with 77 percent withdrawn and 10 percent still under review.[^src-power-demand-forecasts-revised-66709781][^src-data-center-grid-limitations-t-30b0465e]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781",
        "src-data-center-grid-limitations-t-30b0465e"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2000",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781",
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2019",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781",
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "13 percent",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "77 percent",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "10 percent",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781",
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781",
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The timeline from PJM interconnection application to commercial operation has risen from under two years in 2008 to over eight years in 2025.[^src-power-requirements-for-ai-data-c1f60fc1]",
      "source_ids": [
        "src-power-requirements-for-ai-data-c1f60fc1"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2008",
          "found_in": [
            "src-power-requirements-for-ai-data-c1f60fc1"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2025",
          "found_in": [
            "src-power-requirements-for-ai-data-c1f60fc1"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PJM",
          "found_in": [
            "src-power-requirements-for-ai-data-c1f60fc1"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Global data center electricity consumption reached 415 TWh in 2024, representing 1.5% of total global electricity use, according to the International Energy Agency's Energy and AI report (IEA, 2025).",
      "quote_source": "src-power-requirements-for-ai-data-c1f60fc1"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "ERCOT's large-load queue reached 410 GW by April 2026, with 87 percent comprised of data-center requests, exceeding the grid's peak demand by 4.8 times.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "87 percent",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "4.8",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "ERCOT's large-load queue exploded to 410 GW by April 2026\u201487% data centers\u2014exceeding peak demand by 4.8x",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "In Northern Virginia's Data Center Alley, data centers drove 97 percent of 5,250 MW of load growth, with over 30 GW of data center demand queued through 2030 in the Dominion zone alone.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "97 percent",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "In Northern Virginia's Data Center",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Alley",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "MW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Dominion",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "ERCOT's large-load queue exploded to 410 GW by April 2026\u201487% data centers\u2014exceeding peak demand by 4.8x",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Large power transformers above 100 MVA now carry lead times of 120 to 210 weeks, with prices up 79 percent.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "79 percent",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "MVA",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Large power transformers (LPTs >100 MVA) now carry lead times of 120\u2013210 weeks (2.5\u20134+ years), with prices up 79%",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The 2026 Infrastructure Guide similarly reports transformer lead times extending to two to three years due to global manufacturing bottlenecks, and notes that substation capacity frequently caps usable delivery at 250 to 500 MW regardless of generation availability.[^src-the-2026-infrastructure-guide-f419d0b8]",
      "source_ids": [
        "src-the-2026-infrastructure-guide-f419d0b8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-the-2026-infrastructure-guide-f419d0b8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Infrastructure Guide",
          "found_in": [
            "src-the-2026-infrastructure-guide-f419d0b8"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "MW",
          "found_in": [
            "src-the-2026-infrastructure-guide-f419d0b8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Gartner estimates global data center electricity demand will exceed 1,000 TWh by 2026\u2014more than the total consumption of Japan.",
      "quote_source": "src-the-2026-infrastructure-guide-f419d0b8"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The DOE's 2024 National Transmission Planning Study identifies approximately 5,000 miles per year of high-capacity regional transmission as necessary to meet current demand trajectories.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "DOE's",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "National Transmission Planning Study",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The average size of a proposed U.S. data center doubled between 2023 and 2024, from 150 MW to 300 MW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Goldman Sachs suggests that transmission projects can take several years to permit and then several more to build, creating what its research frames as a potential bottleneck for data center growth in regions that are not proactive given lead times.[^src-ai-to-drive-165-increase-in-da-53acaf9f]",
      "source_ids": [
        "src-ai-to-drive-165-increase-in-da-53acaf9f"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "Goldman Sachs",
          "found_in": [
            "src-ai-to-drive-165-increase-in-da-53acaf9f"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Goldman Sachs Research estimates that about $720 billion of grid spending through 2030 may be needed.",
      "quote_source": "src-ai-to-drive-165-increase-in-da-53acaf9f"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Industry analyses place the permitting timeline for regional transmission lines at seven to eleven years.[^src-data-center-grid-limitations-t-30b0465e]",
      "source_ids": [
        "src-data-center-grid-limitations-t-30b0465e"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "interconnection wait times have more than doubled over the past fifteen years, with projects now spending an average of five years in queue before reaching commercial operation",
      "quote_source": "src-data-center-grid-limitations-t-30b0465e"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "One study quantifies U.S. data center water requirements at 697 to 1,451 million gallons per day by 2030 and estimates $10 to $58 billion in new water infrastructure costs.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "money",
          "value": "$10",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "money",
          "value": "$58 billion",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Data centers drove 97% of 5,250 MW load growth in PJM's forecast; the Dominion zone alone has 30+ GW of data center demand queued through 2030",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "One position, reflected in sources discussing liquid cooling efficiency gains and geographic site selection, holds that water challenges are secondary and addressable through direct liquid cooling and geographic diversification.[^src-13-data-center-growth-projecti-dd6338ef][^src-the-2026-infrastructure-guide-f419d0b8]",
      "source_ids": [
        "src-13-data-center-growth-projecti-dd6338ef",
        "src-the-2026-infrastructure-guide-f419d0b8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Global forecasts predict capacity will almost double from 103 gigawatts (GW) to 200 GW by 2030. New capacity additions will reach nearly 100 GW during this five-year period.",
      "quote_source": "src-13-data-center-growth-projecti-dd6338ef"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Elevated interest rates, commodity inflation, and volatility in construction costs are reshaping project economics, while access to inexpensive debt and long-term power contracts is tightening.[^src-the-binding-constraint-current-5d1367ce]",
      "source_ids": [
        "src-the-binding-constraint-current-5d1367ce"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Interconnection queue backlogs, limited substation capacity, and grid congestion are now the dominant determinants of project timelines.",
      "quote_source": "src-the-binding-constraint-current-5d1367ce"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Goldman Sachs estimates approximately $720 billion in grid spending through 2030 may be needed to meet data center power demands.[^src-ai-to-drive-165-increase-in-da-53acaf9f]",
      "source_ids": [
        "src-ai-to-drive-165-increase-in-da-53acaf9f"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-ai-to-drive-165-increase-in-da-53acaf9f"
          ],
          "via": "evidence"
        },
        {
          "kind": "money",
          "value": "$720 billion",
          "found_in": [
            "src-ai-to-drive-165-increase-in-da-53acaf9f"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Goldman Sachs",
          "found_in": [
            "src-ai-to-drive-165-increase-in-da-53acaf9f"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Goldman Sachs Research estimates that about $720 billion of grid spending through 2030 may be needed.",
      "quote_source": "src-ai-to-drive-165-increase-in-da-53acaf9f"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Grid Strategies estimated that faster generation interconnection could have saved consumers as much as $7 billion in that single auction.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "money",
          "value": "$7 billion",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "PJM's forward capacity auction for the 2025/26 delivery year cost consumers $14.7B, up $12.5B from the previous auction's $2.2B price tag. Grid Strategies estimated that, had PJM brought queued resources online more quickly during the preceding years via proactive transmission planning and streamlined interconnection, it could have saved consumers as much as $7B in that single auction, or half of ",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Three sources point to FERC Order 2023, cluster-based study processes, and standardized financial penalties as reforms capable of reducing study timelines from 55 months to under 18 months, framing the problem as solvable through regulatory action within two to three years.[^src-data-center-grid-limitations-t-30b0465e]",
      "source_ids": [
        "src-data-center-grid-limitations-t-30b0465e"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "FERC Order",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        }
      ],
      "quote": "CenterPoint Energy reported a 700% increase in large load interconnection requests between late 2023 and late 2024, growing from 1 gigawatt to 8 gigawatts in a single year.",
      "quote_source": "src-data-center-grid-limitations-t-30b0465e"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "A larger and more independently sourced body of evidence argues that physical infrastructure limitations, transformer shortages, substation saturation, and transmission line construction timelines, persist regardless of procedural reform.[^src-power-demand-forecasts-revised-66709781][^src-550-650-twh-forecast-analysis-b1341866][^src-the-binding-constraint-current-5d1367ce]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781",
        "src-550-650-twh-forecast-analysis-b1341866",
        "src-the-binding-constraint-current-5d1367ce"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Bloom Energy forecasts that 38 percent of data centers will use on-site generation by 2030, up from 13 percent in 2025, representing what the Luminix analysis describes as a 27-times increase in behind-the-meter deployment.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2025",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "38 percent",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "13 percent",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Bloom Energy",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Luminix",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "attribution"
        }
      ],
      "quote": "Data centers drove 97% of 5,250 MW load growth in PJM's forecast; the Dominion zone alone has 30+ GW of data center demand queued through 2030",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "A separate line of analysis finds that behind-the-meter natural gas costs exceed $120 per MWh, well above forward energy prices, and that FERC has restricted large behind-the-meter arrangements on reliability and cost-shifting grounds.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "money",
          "value": "$120",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "FERC",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "PJM's capacity auctions have already priced in scarcity: the 2027/2028 Base Residual Auction cleared at the FERC cap of $333/MW-day but still fell 6,517 MW short of the reliability requirement\u2014the first shortfall exceeding 1 percentage point in history",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The Luminix analysis implies that if 25 to 38 percent of incremental data center load goes behind-the-meter, the grid-served increment drops to roughly 350 to 480 TWh, materially below headline base-case figures.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "38 percent",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Luminix",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Transmission construction timelines of seven to eleven years for permitting alone place new high-capacity lines structurally outside the 2030 window for projects not already in advanced development.[^src-data-center-grid-limitations-t-30b0465e]",
      "source_ids": [
        "src-data-center-grid-limitations-t-30b0465e"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-data-center-grid-limitations-t-30b0465e"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Goldman Sachs Research projects that data center power demand will increase 165% by 2030, requiring approximately $720 billion in grid investment over the decade.",
      "quote_source": "src-data-center-grid-limitations-t-30b0465e"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The 5-year forecast for utility peak load growth has increased by more than a factor of six over three years, from 24 GW in 2022 to 166 GW in 2025.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2022",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2025",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "Yet alternative benchmarks from data center market analysts suggest actual materialization is unlikely to exceed 60 to 65 GW through 2030, implying the utility gross forecasts overstate demand by roughly 25 GW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The average size of a proposed U.S. data center doubled between 2023 and 2024, from 150 MW to 300 MW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "verbatim",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "MW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "verbatim_source": "src-power-demand-forecasts-revised-66709781",
      "verbatim_span": [
        15580,
        15682
      ],
      "quote": "The average size of a proposed U.S. data center doubled between 2023 and 2024, \nfrom 150 MW to 300 MW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "binding-constraints",
      "sentence": "The same Grid Strategies analysis that documents the forecast inflation also documents that only about 65 percent of announced projects may materialize, favoring incumbents with sited power access over speculative builds.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "65 percent",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "only ~65% of announced projects may materialize by 2030 per Grid Strategies benchmarks, favoring incumbents with sited power access over speculative builds",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "**Signal one: speculative queue inflation.** The prior section established that over 2,000 GW of generation and storage capacity sits in U.S. interconnection queues, representing more than the total existing U.S. generation fleet \u2014 a figure the Lawrence Berkeley National Laboratory supplies as the scale anchor for stated demand.[^src-data-centers-and-ai-energy-con-234e7e27]",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "Lawrence Berkeley National Laboratory",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        }
      ],
      "quote": "OpenAI's GPT-4, released in 2023, required an estimated 50 gigawatt-hours (GWh) of electricity for its training run\u2014equivalent to the annual consumption of approximately 40,000 U.S. households or a small city's worth of power.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "As the World Resources Institute reports, utilities are being flooded with early-phase interconnection requests filed at low cost to gauge connection timelines, producing both double-counting of projects and phantom load that will never be built.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [
        {
          "kind": "entity",
          "value": "World Resources Institute",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The WRI analysis further notes that power constraints have already extended data-center construction timelines by 24 to 72 months, and shortages of transformers, switchgear, and gas turbines are compounding the supply-side limitation.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [
        {
          "kind": "entity",
          "value": "WRI",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The Luminix synthesis offers a sharp illustration: ERCOT's large-load queue reached 410 GW by April 2026, with 87% comprised of data-center requests, exceeding peak demand by 4.8x.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "87%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Luminix",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "ERCOT",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "ERCOT's large-load queue exploded to 410 GW by April 2026\u201487% data centers\u2014exceeding peak demand by 4.8x",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "Utility five-year peak-demand forecasts grew from 38 GW in 2023 to 128 GW in 2024, a revision so large it reflects genuine customer commitments rather than speculative noise.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Modeled energy use projections through 2030 range from 200 TWh/year to over 1,050 TWh/year. That highest figure, published by Boston Consulting Group, would represent about a quarter of all U.S. electricity generation in 2023 (which was 4,178 TWh).",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The 2024 LBNL report explicitly avoids the overcounting problem by using a bottom-up model built from installed equipment rather than from interconnection requests, and identifies that methodology as what distinguishes its estimates from utility-reported queue volumes.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Many underlying models on which financial analytics firms build these projections are proprietary or methodologically opaque (Goldman Sachs Research 2024; Jeff Brown 2024; Bloomberg 2024). While not meaningless, historical utility demand forecasts consistently overestimate both peak and average demand (Carvallo et al. 2018).",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "**Signal two: onsite gas generation overbuild.** A news report covering the IEA's *Energy and AI* special report states that natural gas will grow by 175 TWh to meet data-center demand, with much of that growth in the U.S. driven by pro-gas policies and new behind-the-meter generation.[^src-iea-data-center-energy-consump-b5225fe4]",
      "source_ids": [
        "src-iea-data-center-energy-consump-b5225fe4"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "IEA",
          "found_in": [
            "src-iea-data-center-energy-consump-b5225fe4"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The IEA projects that natural gas power will grow by 175TWh to meet data center demand. Much of this growth will be in the US, where the pro-gas policies of the current administration have already led to several major announcements of new gas generation facilities to serve data centers behind the meter.",
      "quote_source": "src-iea-data-center-energy-consump-b5225fe4"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The Luminix report states Bloom Energy forecasts that 38% of data centers will use onsite generation by 2030, up from 13% in 2025, and that RAND estimates 49 GW of behind-the-meter net capacity by 2030 versus 33 GW front-of-meter \u2014 though both figures are stated without supporting evidence in that source.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2025",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "38%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "13%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Luminix",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "Bloom Energy",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "RAND",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Defensible base case: 550\u2013650 TWh (11\u201313% of ~5,000 TWh total US generation). This blends EPRI's medium scenario (596 TWh), LBNL's trend-adjusted high, and Goldman's US-share estimate, assuming hyperscaler dominance (85% of servers), PUE of 1.2\u20131.4, and a roughly even training/inference split.",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "Luminix's own implication is that if 25 to 38 percent of incremental data-center load goes behind the meter, the grid-served increment drops to roughly 350 to 480 TWh, materially less than the headline 550 to 650 TWh base case.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "38 percent",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ]
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The IEA's own report, as summarized by S&P Global, identifies significant uncertainties around AI adoption rates, efficiency improvements, and potential grid bottlenecks that could affect projected demand growth.[^src-spglobal-com-energy-en-news-re-f0591e2d]",
      "source_ids": [
        "src-spglobal-com-energy-en-news-re-f0591e2d"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "IEA's",
          "found_in": [
            "src-spglobal-com-energy-en-news-re-f0591e2d"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "P Global",
          "found_in": [
            "src-spglobal-com-energy-en-news-re-f0591e2d"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-spglobal-com-energy-en-news-re-f0591e2d"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Global electricity demand from data centers is set to more than double to 945 TWh by 2030, equivalent to Japan's current total power consumption, as artificial intelligence drives unprecedented growth in the sector's energy needs, the International Energy Agency said April 10.",
      "quote_source": "src-spglobal-com-energy-en-news-re-f0591e2d"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "Even accepting the report's own range, the lower bound of grid-served demand (roughly 325 TWh) still represents a massive buildout relative to the 176 TWh baseline the 2024 LBNL report establishes for 2023.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "In hindsight, the WRI states, electricity demand was overestimated during that earlier period, with premonitions of power shortages that echo current concerns.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [
        {
          "kind": "entity",
          "value": "WRI",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The *Dev Sustainability* analysis of forecasting methodology reinforces the point: a 2022 review of 258 data-center energy estimates from 46 publications between 2007 and 2021 found only two sources of reliable estimates.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2022",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2007",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2021",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "My 2022 review of 258 data center energy estimates from 46 original publications between 2007 and 2021 showed only two sources of reliable estimates.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "During 2010 to 2016, efficiency gains came from migrating workloads out of inefficient enterprise data centers into efficient hyperscale facilities, a one-time structural shift that kept demand flat despite explosive service growth.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2010",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2016",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "During the early and mid-2010s, a shift from on-premise data centers to colocation or cloud facilities helped enable efficiency improvements that allowed data center electricity use to remain nearly constant at a time when the data center industry grew significantly, with a large expansion of data center services. The efficiency strategies that allowed the industry to avoid increased energy needs ",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The 2024 LBNL report also shows that data-center electricity consumption grew at 18% compounded annually between 2018 and 2023, accelerating sharply from 7% between 2014 and 2018.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2018",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2014",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "18%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "7%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "AI workloads already deploy into the most efficient infrastructure available, as the *Dev Sustainability* analysis observes, so the historical efficiency lever \u2014 workload migration from bad to good infrastructure \u2014 has already been pulled.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "Efficiency gains in AI systems will have to come from better accelerators, better model architectures, higher utilization, better scheduling, and inference optimization.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The major AI labs are already deploying into hyperscale facilities, so new capacity is highly efficient from the start. But that also means the biggest historical efficiency lever - moving workloads out of inefficient enterprise data centers and into modern cloud infrastructure - has already been pulled.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The WRI analysis notes that experts have warned about a rebound effect: even with efficiency improvements, the industry might see increased demand rather than reduced consumption.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [
        {
          "kind": "entity",
          "value": "WRI",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The *Dev Sustainability* analysis describes a workload evolution from chat through reasoning to agents, where software decomposes a single user request into many model calls, tool calls, retrieval steps, and verification loops.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The same source observes that agentic workloads and multi-step task orchestration are also driving CPU demand alongside GPU inference, broadening the energy base of AI beyond the accelerator segment that forecasters track most closely.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "CPU",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The major AI labs are already deploying into hyperscale facilities, so new capacity is highly efficient from the start. But that also means the biggest historical efficiency lever - moving workloads out of inefficient enterprise data centers and into modern cloud infrastructure - has already been pulled.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The *Dev Sustainability* analysis identifies this gap explicitly, noting that agentic workloads and shifting usage patterns complicate energy forecasts and that the source cannot tell us what total workload will be generated as inference scales.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "If inference displaces training as the dominant workload \u2014 Luminix implies a roughly even training-to-inference split in its base case, while other sources suggest inference already dominates[^src-550-650-twh-forecast-analysis-b1341866] \u2014 while algorithmic efficiency simultaneously reduces the compute required per inference task, the bull case faces pressure from two directions at once.",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "attributed",
      "specifics": [
        {
          "kind": "entity",
          "value": "Luminix",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "attribution"
        }
      ]
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The 2024 LBNL report states directly that historical utility demand forecasts have consistently overestimated both peak and average demand, and that many financial-sector models are proprietary or methodologically opaque.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Many underlying models on which financial analytics firms build these projections are proprietary or methodologically opaque (Goldman Sachs Research 2024; Jeff Brown 2024; Bloomberg 2024). While not meaningless, historical utility demand forecasts consistently overestimate both peak and average demand (Carvallo et al. 2018).",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "Against that, the LBNL bottom-up approach itself, which projects a range of 325 to 580 TWh for 2028, requires utilization-rate and GPU-shipment assumptions that are rapidly evolving and difficult to forecast, as the report acknowledges.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "the scenario variations provide a range of total data center energy estimates, with the low and high end of roughly 325 and 580 TWh in 2028, as shown in Figure ES-1. Assuming an average capacity utilization rate of 50%, this annual energy use range would translate to a total power demand for data centers between 74 and 132 GW. This annual energy use also represents 6.7% to 12.0% of total U.S. elec",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The IEA's 945 TWh global projection for 2030 is reported without confidence intervals or scenario probabilities in the secondary accounts available.[^src-iea-data-center-energy-consump-b5225fe4][^src-spglobal-com-energy-en-news-re-f0591e2d]",
      "source_ids": [
        "src-iea-data-center-energy-consump-b5225fe4",
        "src-spglobal-com-energy-en-news-re-f0591e2d"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-iea-data-center-energy-consump-b5225fe4",
            "src-spglobal-com-energy-en-news-re-f0591e2d"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "IEA's",
          "found_in": [
            "src-spglobal-com-energy-en-news-re-f0591e2d"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The Energy and AI report projects that data center consumption globally will grow to 945TWh per year by 2030, from 415TWh in 2024.",
      "quote_source": "src-iea-data-center-energy-consump-b5225fe4"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The *Dev Sustainability* author summarizes the analytical situation fairly: extreme high-end scenarios are useful for getting an estimate quoted in the press rather than as realistic projections of actual consumption, while the economic incentive to improve efficiency will eventually manifest in actual consumption.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "overbuild-bubble",
      "sentence": "The LBNL report's 18% compound annual growth rate between 2018 and 2023 is not speculative \u2014 it is measured against actual installed equipment.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2018",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "18%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "The LBNL baseline established earlier \u2014 176 TWh in 2023, with projections ranging from 325 to 580 TWh by 2028 \u2014 anchors the scale of that prize.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "Utility five-year peak load growth forecasts have risen more than six-fold over three years, from 24 GW to 166 GW, with data centers accounting for roughly 55% of that growth.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "55%",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "The analysis notes that some experts have already raised the alarm about current rate structures and cost allocation methods leading to the possibility of other residential or commercial consumers paying for infrastructure used only to service new data centers.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Modeled energy use projections through 2030 range from 200 TWh/year to over 1,050 TWh/year. That highest figure, published by Boston Consulting Group, would represent about a quarter of all U.S. electricity generation in 2023 (which was 4,178 TWh).",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "Grid Strategies estimated that faster interconnection of queued resources could have saved consumers as much as $7 billion in that single auction.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "money",
          "value": "$7 billion",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "PJM's forward capacity auction for the 2025/26 delivery year cost consumers $14.7B, up $12.5B from the previous auction's $2.2B price tag. Grid Strategies estimated that, had PJM brought queued resources online more quickly during the preceding years via proactive transmission planning and streamlined interconnection, it could have saved consumers as much as $7B in that single auction, or half of ",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "PJM's 2027/2028 Base Residual Auction cleared at the FERC cap of $333 per MW-day but still fell 6,517 MW short of the reliability requirement \u2014 the first shortfall exceeding one percentage point in PJM history.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2027",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "money",
          "value": "$333",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Base Residual Auction",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "FERC",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "MW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PJM",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "PJM's capacity auctions have already priced in scarcity: the 2027/2028 Base Residual Auction cleared at the FERC cap of $333/MW-day but still fell 6,517 MW short of the reliability requirement\u2014the first shortfall exceeding 1 percentage point in history",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "In retrospect, electricity demand was overestimated, with concerns about power shortages echoing current anxieties.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Modeled energy use projections through 2030 range from 200 TWh/year to over 1,050 TWh/year. That highest figure, published by Boston Consulting Group, would represent about a quarter of all U.S. electricity generation in 2023 (which was 4,178 TWh).",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "The WRI explicitly notes that historical utility demand forecasts have consistently overestimated both peak and average demand \u2014 a finding LBNL's 2024 report echoes directly.[^src-eta-publications-lbl-gov-sites-f3f409d8][^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8",
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "WRI",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Many underlying models on which financial analytics firms build these projections are proprietary or methodologically opaque (Goldman Sachs Research 2024; Jeff Brown 2024; Bloomberg 2024). While not meaningless, historical utility demand forecasts consistently overestimate both peak and average demand (Carvallo et al. 2018).",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "Project Stargate, announced in January with $500 billion in stated AI data-center spending by 2029, had experienced a slower than expected start and had announced plans for only one small data center to open by year-end.[^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2029",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "money",
          "value": "$500 billion",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Project Stargate",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-the-challenge-of-forecasting-e-6b943a17"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Power availability has already emerged as a limiting factor for many data center developers, with one analysis finding that power constraints were extending data center construction timelines by 24 to 72 months. Shortages of infrastructure components like transformers, switchgears and gas turbines are only compounding this issue.",
      "quote_source": "src-the-challenge-of-forecasting-e-6b943a17"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "The LBNL report observes that the underlying models on which financial analytics firms build their projections are proprietary or methodologically opaque.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ]
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "On the behind-the-meter side, a projection that 38% of data centers will use on-site generation by 2030, up from 13% in 2025, and RAND's estimate that 49 GW of behind-the-meter capacity will exceed 33 GW front-of-meter by 2030, both carry the weight of assertions without independent verification in the available record.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2025",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "38%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "13%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "RAND",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Defensible base case: 550\u2013650 TWh (11\u201313% of ~5,000 TWh total US generation). This blends EPRI's medium scenario (596 TWh), LBNL's trend-adjusted high, and Goldman's US-share estimate, assuming hyperscaler dominance (85% of servers), PUE of 1.2\u20131.4, and a roughly even training/inference split.",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "LBNL's bottom-up equipment model builds from commercial equipment shipment data, utilization assumptions, and PUE trajectories rather than from project-pipeline optimism or workload extrapolation.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-ai-data-center-energy-in-2026-fb073872"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "That methodology avoids the overestimation that can arise from tracking data-center load for projects that have not yet selected a power provider.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "Large power transformers now carry lead times of 120 to 210 weeks with prices up 79%.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "79%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Large power transformers (LPTs >100 MVA) now carry lead times of 120\u2013210 weeks (2.5\u20134+ years), with prices up 79%",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "The average size of a proposed US data center doubled from 150 MW to 300 MW between 2023 and 2024.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "US",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "MW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "By 2030, forecasts indicate that total electricity use will increase by 32%. The higher growth rate for electricity use likely reflects high load factors of data centers as well as forecast changes in off-peak energy use by other customers.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "Grid interconnection timelines from PJM application to commercial operation have risen from under two years in 2008 to over eight years in 2025.[^src-power-requirements-for-ai-data-c1f60fc1]",
      "source_ids": [
        "src-power-requirements-for-ai-data-c1f60fc1"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2008",
          "found_in": [
            "src-power-requirements-for-ai-data-c1f60fc1"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2025",
          "found_in": [
            "src-power-requirements-for-ai-data-c1f60fc1"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PJM",
          "found_in": [
            "src-power-requirements-for-ai-data-c1f60fc1"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Global data center electricity consumption reached 415 TWh in 2024, representing 1.5% of total global electricity use, according to the International Energy Agency's Energy and AI report (IEA, 2025).",
      "quote_source": "src-power-requirements-for-ai-data-c1f60fc1"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "Cleanview tracked on the order of 50-plus GW of announced behind-the-meter or co-located generation associated with data center projects, with most identifiable equipment being natural gas-based.[^src-data-centers-need-the-grid-so--0cdfb18c]",
      "source_ids": [
        "src-data-centers-need-the-grid-so--0cdfb18c"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-data-centers-need-the-grid-so--0cdfb18c"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Utilities are already acknowledging that their forecasts were wrong. A Grid Strategies compilation found that, over the last three years, utilities' five-year peak load growth forecasts jumped from roughly 24 GW to about 166 GW, with data centers a major driver.",
      "quote_source": "src-data-centers-need-the-grid-so--0cdfb18c"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "The EIA projects 3.3 GW of combined-cycle natural gas generation online in 2026, another 3.3 GW in 2027, and 10.6 GW in 2028.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2027",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "3.3",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "10.6",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "EIA",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "ERCOT's large-load queue exploded to 410 GW by April 2026\u201487% data centers\u2014exceeding peak demand by 4.8x",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "Sources documenting physical infrastructure constraints \u2014 transformer lead times, substation saturation, and transmission construction timelines that run 7 to 11 years for permitting alone \u2014 represent three independent lines.[^src-550-650-twh-forecast-analysis-b1341866][^src-data-centers-need-the-grid-so--0cdfb18c][^src-power-requirements-for-ai-data-c1f60fc1]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866",
        "src-data-centers-need-the-grid-so--0cdfb18c",
        "src-power-requirements-for-ai-data-c1f60fc1"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "The range of published forecasts is extraordinary: from IEA's ~425 TWh (8% of US total) to BCG's ~970 TWh (19%), a spread so wide it's functionally useless for infrastructure planning. The disagreement is not noise\u2014it reflects genuinely different modeling philosophies colliding with a genuinely unprecedented demand driver.",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "AI is being deployed into the most efficient infrastructure available, so further gains must come from better accelerators, model architectures, utilization scheduling, and inference optimization rather than from infrastructure consolidation.[^src-ai-data-center-energy-in-2026-fb073872]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "It represents one analytical voice, not independent corroboration.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "winners-payers-effects",
      "sentence": "That range implies the grid-served increment could fall to roughly 350 to 480 TWh if 25 to 38% of incremental load goes behind-the-meter \u2014 still a substantial buildout, but materially below headline figures.[^src-550-650-twh-forecast-analysis-b1341866]",
      "source_ids": [
        "src-550-650-twh-forecast-analysis-b1341866"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "38%",
          "found_in": [
            "src-550-650-twh-forecast-analysis-b1341866"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Bloom Energy forecasts that 38% of data centers will use on-site generation by 2030 (up from 13% in 2025), with 33% operating fully off-grid\u2014a 27x increase",
      "quote_source": "src-550-650-twh-forecast-analysis-b1341866"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The LBNL 2024 bottom-up model projects 325 to 580 TWh by 2028, representing 6.7% to 12.0% of forecasted US electricity consumption, driven by AI server shipments, GPU power ramps running at 60% to 80% of rated capacity, and PUE declining toward roughly 1.4 through liquid cooling and hyperscale consolidation.[^src-eta-publications-lbl-gov-sites-f3f409d8][^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8",
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "6.7%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "12.0%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "60%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "80%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.4",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "US",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8",
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The LBNL team reports a compound annual growth rate of approximately 7% from 2014 to 2018, accelerating to 18% between 2018 and 2023, with the 2023-to-2028 band ranging from 13% to 27%.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2014",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2018",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "7%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "18%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "13%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "27%",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "A synthesis that extrapolates LBNL's reported CAGR band forward to 2030 yields an approximate range of 450 to 850 TWh.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "CAGR",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The EPRI state-level pipeline analysis, which aggregates operational, under-construction, and announced projects under low, medium, and high realization scenarios, projects 380 to 790 TWh by 2030, representing 9% to 17% of US electricity use, converting IT capacity estimates of 56 to 132 GW to energy through implicit load factors and PUE assumptions.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "9%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "17%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "EPRI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "US",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "IT",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "That pipeline estimate represents a 60% upward revision from EPRI's 2024 forecast, driven by 18 months of accelerated AI infrastructure announcements.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "60%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "EPRI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Goldman Sachs projects approximately 810 TWh of US data-center electricity demand by 2030, at 14% to 16% of US total, based on a 220% global growth scenario with a 60% US share of incremental demand driven by hyperscaler capital expenditure surges exceeding $300 billion.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "money",
          "value": "$300 billion",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "14%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "16%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "220%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "60%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Goldman Sachs",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "US",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Blending the EPRI medium scenario, McKinsey's 606 TWh workload-driven estimate, and the LBNL high-end figure, and weighting for hyperscaler dominance at roughly 85% of servers, a PUE range of 1.2 to 1.4, and an approximate 50/50 training-to-inference split, yields a defensible base case of 550 to 650 TWh by 2030.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "85%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.2",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.4",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "EPRI",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "LBNL",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "A sensitivity analysis attached to this synthesis indicates potential variance of plus 20% if inference workloads exceed training, minus 15% if GPU efficiency and PUE below 1.2 are achieved, and plus or minus 10% from project realization risk.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "20%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "15%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "10%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "number",
          "value": "1.2",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GPU",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "PUE",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "At the global level, the International Energy Agency projects that data center electricity consumption could double by 2030, potentially reaching 600 to 800 TWh annually.[^src-data-centers-and-ai-energy-con-234e7e27]",
      "source_ids": [
        "src-data-centers-and-ai-energy-con-234e7e27"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "International Energy Agency",
          "found_in": [
            "src-data-centers-and-ai-energy-con-234e7e27"
          ],
          "via": "evidence"
        }
      ],
      "quote": "The International Energy Agency (IEA) projects that data center electricity consumption could double by 2030, potentially reaching 600-800 TWh annually.",
      "quote_source": "src-data-centers-and-ai-energy-con-234e7e27"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The Belfer Center reports that institutional projections of data center electricity demand range from about 200 TWh to over 1,000 TWh by 2030, a spread that itself confirms the range cannot be collapsed to a point estimate.[^src-ai-data-centers-and-the-u-s-el-73756c3b]",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Belfer Center",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "attribution"
        }
      ],
      "quote": "institutional projections of data center electricity demand range from about 200 TWh to over 1,000 TWh by 2030, according to the World Resources Institute. This range complicates medium-to-long term grid planning",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Two independent analytical lines argue that efficiency gains from the 2010s, which kept data center demand nearly flat despite surging service demand, drew primarily on a one-time structural shift from on-premises to hyperscale facilities.[^src-ai-data-center-energy-in-2026-fb073872][^src-the-challenge-of-forecasting-e-6b943a17]",
      "source_ids": [
        "src-ai-data-center-energy-in-2026-fb073872",
        "src-the-challenge-of-forecasting-e-6b943a17"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Cloud migration, better utilization, virtualization, improved PUE, and more efficient equipment meant digital services could grow without electricity demand rising nearly as fast.",
      "quote_source": "src-ai-data-center-energy-in-2026-fb073872"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The IEA, by contrast, has modeled 175 GW of transmission capacity potentially unlockable through AI-based grid management and 13 exajoules of energy savings through AI-enabled industrial optimization, figures that could theoretically offset incremental data center load.[^src-iea-org-reports-key-questions--e71ee1cf][^src-iea-org-reports-energy-and-ai--7a179a48]",
      "source_ids": [
        "src-iea-org-reports-key-questions--e71ee1cf",
        "src-iea-org-reports-energy-and-ai--7a179a48"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "IEA",
          "found_in": [
            "src-iea-org-reports-key-questions--e71ee1cf",
            "src-iea-org-reports-energy-and-ai--7a179a48"
          ],
          "via": "attribution"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-iea-org-reports-energy-and-ai--7a179a48"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "AI",
          "found_in": [
            "src-iea-org-reports-key-questions--e71ee1cf",
            "src-iea-org-reports-energy-and-ai--7a179a48"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past year, energy and technology supply chains have become further strained. Trade restrictions have targeted key components going into data centres, such as the critical minerals needed for advanced power electronics, or the batteries and battery components needed to smooth AI loads.",
      "quote_source": "src-iea-org-reports-key-questions--e71ee1cf"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The case that FERC Order 2023 and grid-enhancing technologies can reduce study timelines from 55 months to under 18 months draws on one analytical line echoed across three sources.[^src-novoco-com-notes-from-novograd-429dcfc3][^src-the-interconnection-queue-cont-6a97e40c][^src-lawrence-berkeley-lab-2025-int-e4901690]",
      "source_ids": [
        "src-novoco-com-notes-from-novograd-429dcfc3",
        "src-the-interconnection-queue-cont-6a97e40c",
        "src-lawrence-berkeley-lab-2025-int-e4901690"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-novoco-com-notes-from-novograd-429dcfc3",
            "src-the-interconnection-queue-cont-6a97e40c",
            "src-lawrence-berkeley-lab-2025-int-e4901690"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "FERC Order",
          "found_in": [
            "src-novoco-com-notes-from-novograd-429dcfc3",
            "src-the-interconnection-queue-cont-6a97e40c",
            "src-lawrence-berkeley-lab-2025-int-e4901690"
          ],
          "via": "evidence"
        }
      ],
      "quote": "As of the end of 2023, an astonishing 2,600 GWs (2.6 terawatts) of proposed generation and storage projects were waiting for grid access, with typical wait times stretching to five years or more.",
      "quote_source": "src-novoco-com-notes-from-novograd-429dcfc3"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The case that physical infrastructure constraints will persist regardless of procedural reform draws on three independent lines: transformer lead times exceeding 200 weeks, substation capacity exhaustion, and transmission permitting timelines of 7 to 11 years.[^src-power-demand-forecasts-revised-66709781][^src-the-binding-constraint-current-5d1367ce][^src-data-center-grid-limitations-t-30b0465e]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781",
        "src-the-binding-constraint-current-5d1367ce",
        "src-data-center-grid-limitations-t-30b0465e"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The 5-year utility peak load growth forecast has risen by more than a factor of six over three years, from 24 GW to 166 GW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Data centers account for roughly 55% of that forecasted growth, or approximately 90 GW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "percent",
          "value": "55%",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Alternative benchmarks from data center market analysts suggest actual demand growth is unlikely to exceed 60 to 65 GW through 2030, implying utility forecasts collectively overstate data center demand by roughly 25 GW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Only approximately 65% of announced projects may materialize by 2030, per Grid Strategies benchmarks.[^src-research-and-synthesize-the-le-bef42f4b]",
      "source_ids": [
        "src-research-and-synthesize-the-le-bef42f4b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2030",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "65%",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Grid Strategies",
          "found_in": [
            "src-research-and-synthesize-the-le-bef42f4b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "EPRI's state-level pipeline tracking reveals data centers could drive 9-17% of US electricity use by 2030: they aggregate operational, under-construction, and announced projects into low/medium/high scenarios based on realization rates (e.g., low assumes 25% advanced planning online; high adds 30% early planning), converting IT capacity (56-132 GW) to energy via implicit load factors/PUE, yielding",
      "quote_source": "src-research-and-synthesize-the-le-bef42f4b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Of sixteen GW-scale data centers scheduled to come online in 2026 and 2027 with aggregate demand of nearly 30 GW, nearly all hold initial permits and around half have begun construction.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2026",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2027",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "GW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Over the past three years, the 5-year forecast of utility peak load growth has increased by more than a factor of six, from 24 GW to 166 GW.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "LBNL argues that bottom-up equipment models are more defensible than pipeline extrapolation methods, and documents that historical utility demand forecasts have consistently overestimated both peak and average demand.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Counterarguments exist: modern forecasts represent a real methodological improvement over earlier work, given that a review of 258 data center energy consumption estimates found systematic methodological defects across the field, particularly regarding data availability and transparency.[^src-ai-data-centers-and-the-u-s-el-73756c3b]",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 terawatt hours (TWh) in 2023 (or, about 4.4% of total U.S. electricity consumption) to between 325-580 TWh (6.7-12.0%) by 2028",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The reconciled position is that modern estimates are more rigorous, but that the uncertainty range from 325 to 580 TWh by 2028 reflects genuine epistemological limits, not analytical failure.[^src-eta-publications-lbl-gov-sites-f3f409d8]",
      "source_ids": [
        "src-eta-publications-lbl-gov-sites-f3f409d8"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2028",
          "found_in": [
            "src-eta-publications-lbl-gov-sites-f3f409d8"
          ],
          "via": "evidence"
        }
      ],
      "quote": "the scenario variations provide a range of total data center energy estimates, with the low and high end of roughly 325 and 580 TWh in 2028, as shown in Figure ES-1. Assuming an average capacity utilization rate of 50%, this annual energy use range would translate to a total power demand for data centers between 74 and 132 GW. This annual energy use also represents 6.7% to 12.0% of total U.S. elec",
      "quote_source": "src-eta-publications-lbl-gov-sites-f3f409d8"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "Amazon, Microsoft, Google, and Meta collectively spent over $200 billion on capital expenditure in 2024, a 62% year-over-year increase from 2023.[^src-ai-data-centers-and-the-u-s-el-73756c3b]",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        },
        {
          "kind": "money",
          "value": "$200 billion",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        },
        {
          "kind": "percent",
          "value": "62%",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Microsoft",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Google",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "Meta",
          "found_in": [
            "src-ai-data-centers-and-the-u-s-el-73756c3b"
          ],
          "via": "evidence"
        }
      ],
      "quote": "Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 terawatt hours (TWh) in 2023 (or, about 4.4% of total U.S. electricity consumption) to between 325-580 TWh (6.7-12.0%) by 2028",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The average size of a proposed US data center doubled between 2023 and 2024, from 150 MW to 300 MW.[^src-power-demand-forecasts-revised-66709781]",
      "source_ids": [
        "src-power-demand-forecasts-revised-66709781"
      ],
      "entailment_class": "grounded",
      "specifics": [
        {
          "kind": "year",
          "value": "2023",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "year",
          "value": "2024",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "US",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        },
        {
          "kind": "entity",
          "value": "MW",
          "found_in": [
            "src-power-demand-forecasts-revised-66709781"
          ],
          "via": "evidence"
        }
      ],
      "quote": "By 2030, forecasts indicate that total electricity use will increase by 32%. The higher growth rate for electricity use likely reflects high load factors of data centers as well as forecast changes in off-peak energy use by other customers.",
      "quote_source": "src-power-demand-forecasts-revised-66709781"
    },
    {
      "section_id": "synthesis-investable-view",
      "sentence": "The scenario in which demand falls materially short exposes ratepayers to stranded costs from unnecessary infrastructure investment; utilities with the largest demand-driven build-out programs carry the most concentrated bear-case risk.[^src-ai-data-centers-and-the-u-s-el-73756c3b]",
      "source_ids": [
        "src-ai-data-centers-and-the-u-s-el-73756c3b"
      ],
      "entailment_class": "attributed",
      "specifics": [],
      "quote": "Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 terawatt hours (TWh) in 2023 (or, about 4.4% of total U.S. electricity consumption) to between 325-580 TWh (6.7-12.0%) by 2028",
      "quote_source": "src-ai-data-centers-and-the-u-s-el-73756c3b"
    }
  ]
}