{
  "schemaVersion": "1.0",
  "kind": "nilu.article",
  "id": "rede-eletrica-freio-crescimento",
  "slug": "electric-grid-growth-constraint",
  "locale": "en",
  "sourceLocale": "pt-br",
  "localizationStatus": "translated",
  "title": "Electric grid as a growth constraint",
  "summary": "M1–M4 map of the grid lot, labeled IEA/World Bank KPIs, and the gen≠load, queue≠COD, Europe≠one-grid locks.",
  "author": "NiLu",
  "publishedAt": "2026-08-09T00:00:00.000Z",
  "updatedAt": "2026-08-09T00:00:00.000Z",
  "canonicalUrl": "https://blog.nilu.pro/en/blog/electric-grid-growth-constraint/",
  "representations": {
    "html": "https://blog.nilu.pro/en/blog/electric-grid-growth-constraint/",
    "markdown": "https://blog.nilu.pro/en/blog/electric-grid-growth-constraint.md",
    "json": "https://blog.nilu.pro/en/blog/electric-grid-growth-constraint.json"
  },
  "bodyMarkdown": "# Electric grid as a growth constraint\n\nM1–M4 map of the grid lot, labeled IEA/World Bank KPIs, and the gen≠load, queue≠COD, Europe≠one-grid locks.\n\nAuthor: NiLu\nLocale: en\nPublished: 2026-08-09\nUpdated: 2026-08-09\nCanonical: https://blog.nilu.pro/en/blog/electric-grid-growth-constraint/\nMarkdown: https://blog.nilu.pro/en/blog/electric-grid-growth-constraint.md\nJSON: https://blog.nilu.pro/en/blog/electric-grid-growth-constraint.json\n\n## Article summary\n\n- The lot reads \"grid as a growth constraint\" through four chained mechanisms — physics (M1), queue and reinforcement (M2), who pays and decides (M3), and how this plays out in Brazil, the US, and Europe (M4).\n- Honest KPIs from the IEA and the World Bank — investment, queue stock, build lead times, AI demand — always come with a vintage and a scope, never as an average crossing jurisdictions.\n- Three technical locks: generation queue ≠ load process; MW stuck in the queue ≠ COD/energization; Europe is not one grid nor a single queue.\n- Concession terms are a dated legal calendar, not a judgment on any utility; ongoing disputes appear only as attributed perspectives, detailed in the specialist dossiers.\n\n## Contents\n\n- [Why the grid became the constraint](#grid-as-the-constraint)\n- [The map: four mechanisms (M1–M4)](#the-map-four-mechanisms-m1-m4)\n- [Honest KPIs: investment, queue, lead times](#honest-kpis)\n- [Demand and AI in TWh, not MW](#demand-and-ai-in-twh)\n- [Locks that cannot be crossed](#locks-that-cannot-be-crossed)\n- [Open disputes, no verdict](#open-disputes)\n- [What we still don't know](#uncertainties)\n\n<a id=\"grid-as-the-constraint\"></a>\n\n## Why the grid became the constraint\n\nGrowing with electricity — industry, data centres, transport and heat electrification — depends on the grid delivering power exactly where the load is. The IEA describes this as a bottleneck: insufficient grid capacity blocks the connection of generation, demand, and storage, and connection queues are at record levels. [[1]](#ref-1) Grid investment remains near USD 400 billion per year — trailing the pace of generation, near USD 1 trillion, and of accelerating demand. [[1]](#ref-1) [[2]](#ref-2)\n\nThis isn't a single number saying \"the world is behind.\" It's a map of mechanisms that plays out differently in each jurisdiction — and that map is what this dossier organizes before any local detail.\n\n<a id=\"the-map-four-mechanisms-m1-m4\"></a>\n\n## The map: four mechanisms (M1–M4)\n\nThe NiLu lot on the electric grid reads the constraint through four chained layers:\n\n- **M1 — Physics.** Generation, transmission, distribution, and load form a chain balanced in real time. \"Capacity,\" in lay usage, blends distinct ideas — maximum generation, line limit, contingency margin, queue — and delivery congestion at one point is not the same thing as national energy scarcity. Layers and units, in detail, are in How an electric grid works.\n- **M2 — Queue, study, reinforcement.** Connecting something to the grid is a process, not a fixed MW threshold: request → study → access opinion/reinforcement. The generation queue follows a different track from the load queue, and megawatts stuck in a queue are not equivalent to COD (commercial operation date). Detail in Capacity, interconnection queues, and reinforcement.\n- **M3 — Who decides, who pays.** Cost allocation methods (socialized, beneficiary, participant, CBCA) and institutional roles — regulator, operator, granting authority — are mechanisms documented per jurisdiction; concession terms are a legal calendar, not a moral judgment. Detail in Who pays and who decides in grid expansion.\n- **M4 — Where this happens.** The same M1–M3 play out in different ways: Brazil (the SIN + local evacuation constraints assessed by the EPE), the United States (FERC/NERC and ISOs/RTOs, with a generation queue and a separate track for large loads), and Europe (ENTSO-E and ACER coordinating national regulators — not one single synchronous grid nor a pan-European queue). Detail in the Brazil, United States, and Europe articles.\n\n<a id=\"honest-kpis\"></a>\n\n## Honest KPIs: investment, queue, lead times\n\nThree families of numbers deserve a seat at any manager's table — always labeled, never as a single index.\n\n**Investment.** Grids run near USD 400 billion per year; generation, near USD 1 trillion — in 2025, the electricity sector as a whole totaled approximately USD 1,5 trillion. The IEA describes the grid as \"struggling to keep pace\" with demand and renewables, and advocates for grid spending moving toward parity with generation spending — held back by permitting, transformer and cable supply chains, and, especially in developing economies, weak utility finances. [[2]](#ref-2) For 2030, the IEA itself estimates that grid investment needs to grow by about 50% over the current level. [[1]](#ref-1)\n\n**Queue stock — with method, not without it.** Electricity 2026 records more than 2.500 GW of projects — renewables, large loads, and storage — stuck in connection queues worldwide, an indicative figure for 2025. [[1]](#ref-1) An earlier report, from October 2023, measured at least 3.000 GW of renewables awaiting connection (1.500 GW at an advanced stage), covering countries that together account for about half of global wind+solar capacity — which suggests the real backlog is larger than the reported figure. [[4]](#ref-4) These are two different methods and scopes: one is renewables only; the other adds large loads and storage. Do not add them together, do not average them.\n\n**Build lead times.** Building new grid infrastructure typically takes 5 to 15 years between planning, permitting, and completion — against 1 to 5 years for renewables, 1 to 3 years for data centres, and 1 to 2 years for electric-vehicle charging. [[1]](#ref-1) The same IEA points to partial relief: non-firm connections combined with grid-enhancing technologies could free up between 1.200 and 1.600 GW of hosting capacity for advanced-stage projects — the two levers are not additive with each other. [[1]](#ref-1)\n\nIn 2023, the IEA already described grid investment as stagnant near USD 300 billion/year, needing to nearly double by 2030 (to more than USD 600 billion/year), with a build scale equivalent to adding or refurbishing about 80 million km of grid by 2040. [[4]](#ref-4) The same report models a delay scenario (the \"Grid Delay Case\") that would result in approximately +58 Gt of cumulative CO₂ by 2050 versus a goals-aligned trajectory — a scenario outcome reported by the IEA itself, not a verdict by NiLu about any actor. [[4]](#ref-4)\n\nOn the financial side, the World Bank documents that, in the UPBEAT database (more than 180 utilities in more than 90 countries), fewer than 40% recover enough revenue to cover operating costs and debt service — under-recovery linked to underinvestment in maintenance, upgrades, and expansion; decarbonizing a hypothetical power mix by 2050 could raise the required transmission grid intensity by about 30%. [[6]](#ref-6) When wheeling tariffs are missing or inconsistent, the World Bank/ESMAP proposes the LCOT (Levelized Cost of Transmission, in USD per kWh transported) as an analytical benchmark — not as a universal tariff nor as a cost-allocation rule, which continues to be decided case by case and is detailed in Who pays and who decides in grid expansion. [[5]](#ref-5)\n\n<a id=\"demand-and-ai-in-twh\"></a>\n\n## Demand and AI in TWh, not MW\n\nData centres and AI enter the debate as if every TWh consumed corresponded to one MW of connection queue — they don't. The IEA estimates that data centres consumed about 415 TWh in 2024 (≈1,5% of global electricity), with a Base Case of approximately 945 TWh in 2030 (≈3% of the global total); the projected growth, of about 15% per year between 2024 and 2030, would be more than 4 times the electricity-consumption growth of other sectors — with accelerated (AI-oriented) servers growing at around 30% per year against about 9% per year for conventional servers. [[7]](#ref-7)\n\nIn the broader picture, global electricity consumption is on a trajectory toward more than 29.000 TWh in 2026; in the United States, data centres consumed about 180 TWh in 2024, with a projection of a further 240 TWh added by 2030 relative to 2024. [[8]](#ref-8) Demand is also accelerating more broadly: the IEA projects growth of 3,3% (2025) and 3,7% (2026) in global electricity demand — among the highest rates of the past decade, against an average of 2,6% over 2015–2023 —, with electrification, industry, air conditioning, and data centres named as drivers. [[3]](#ref-3) In the United States, estimated growth is 2,3% (2025) / 2,2% (2026), with data centres as a driver; in the European Union, the pace is more modest, at 1,1% (2025) / 1,5% (2026). [[3]](#ref-3)\n\nThese numbers describe energy consumed (TWh), not power stuck in an interconnection queue (MW), nor peak demand, nor COD. Jurisdiction-level detail — including the specific large-load tracks in the US and the national hubs in Europe — is covered in the United States and Europe articles.\n\n<a id=\"locks-that-cannot-be-crossed\"></a>\n\n## Locks that cannot be crossed\n\nFour technical locks organize the reading of this lot and must not be crossed in any text about the grid as a growth constraint:\n\n1. **Generation queue ≠ load process.** Connecting a generation project and connecting a large load (data centre, industry) follow separate study tracks, even within the same operator — mixing the two produces meaningless numbers.\n2. **MW in the queue ≠ COD.** A project \"in the queue\" can be at any stage between the initial request and energization; MW queue stock is not a forecast of capacity that will enter operation, nor does it indicate when.\n3. **Europe is not one grid nor a single queue.** ENTSO-E coordinates network codes and planning (TYNDP, ERAA), and ACER coordinates national regulators — but each country keeps its own connection process and, in some cases, its own large-load waitlist.\n4. **Concession terms are a dated legal fact.** Renewal, extension, or termination of a distribution or transmission concession follows a calendar set by law, contract, or tender — it is information verifiable by date, not a verdict on any utility's management.\n\nThese four locks apply equally to whoever builds a CapEx risk map and whoever fact-checks a headline: naming the mechanism (M1–M4), the jurisdiction, and the metric (TWh vs. MW; request vs. COD; local vs. national) avoids most reading errors.\n\n<a id=\"open-disputes\"></a>\n\n## Open disputes, no verdict\n\nPart of the public debate about the grid as a constraint involves disputes still in progress — and this dossier treats them as attributed perspectives, not as a conclusion of the lot. Coverage in the Brazilian press (Valor, G1, Poder360, iNFRA) tracks an ANEEL proceeding involving the distribution utility Enel in São Paulo, on the concession forfeiture/renewal axis; the legal mechanism for extension and suspension is detailed in Who pays and who decides in grid expansion and in Electric grid in Brazil, and the press facts remain under observation pending official documentation. In the United States, FERC communications around the large-load show-cause orders, in June 2026, emphasize ratepayer protection and the states' role — an institutional voice from the regulatory process, not a verdict on hyperscalers versus consumers; detail in Electric grid in the United States. In Europe, national regulators and operators — such as CRU/EirGrid in Ireland and the TenneT/ACM cluster in the Netherlands — publish their own policies and waitlists for data centres, which reinforces that there is no \"EU data centre policy\" nor a single pan-European queue dataset; detail in Electric grid in Europe.\n\n<a id=\"uncertainties\"></a>\n\n## What we still don't know\n\n- There is no harmonized definition of \"connection queue\" that allows comparing the 2023 stock (renewables only) with the 2026 stock (renewables + large loads + storage) without relying on manual labels.\n- There is no comparable public dataset for the load queue (load interconnection) in the US or at the pan-European level, distinct from portals like Queued Up/GI or the national waitlists of Ireland and the Netherlands.\n- It is not clear how the World Bank's LCOT connects, in practice, to cost-allocation methods such as Order 1000/1920, RAP→TUST, or CBCA without being read as a universal tariff.\n- There is still no short, stable package of \"grid as a constraint\" KPIs for dual reading (manager/technical) that doesn't turn into an invented composite index.\n\n## References\n\n1. <a id=\"ref-1\"></a>[IEA — Grids, Electricity 2026](https://www.iea.org/reports/electricity-2026/grids). IEA. Accessed August 09, 2026.\n2. <a id=\"ref-2\"></a>[IEA — World Energy Investment 2025, Executive summary](https://www.iea.org/reports/world-energy-investment-2025/executive-summary). IEA. Accessed August 09, 2026.\n3. <a id=\"ref-3\"></a>[IEA — Electricity Mid-Year Update 2025, Executive summary](https://www.iea.org/reports/electricity-mid-year-update-2025/executive-summary). IEA. Accessed August 09, 2026.\n4. <a id=\"ref-4\"></a>[IEA — Electricity Grids and Secure Energy Transitions](https://www.iea.org/reports/electricity-grids-and-secure-energy-transitions). IEA. Accessed August 09, 2026.\n5. <a id=\"ref-5\"></a>[World Bank / ESMAP — Understanding the Cost of Transmission Infrastructure](https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099040926152055780). World Bank. Accessed August 09, 2026.\n6. <a id=\"ref-6\"></a>[World Bank — The Critical Link: Empowering Utilities for the Energy Transition](https://documents1.worldbank.org/curated/en/099052124103035656/pdf/P500450-a30f897d-bc13-42b4-88cd-61d27a55a930.pdf). World Bank. Accessed August 09, 2026.\n7. <a id=\"ref-7\"></a>[IEA — Energy demand from AI, Energy and AI](https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai). IEA. Accessed August 09, 2026.\n8. <a id=\"ref-8\"></a>[IEA — Demand chapter, Electricity Mid-Year Update 2025](https://www.iea.org/reports/electricity-mid-year-update-2025/demand-global-electricity-use-to-grow-strongly-in-2025-and-2026). IEA. Accessed August 09, 2026.\n"
}