The Information Machine

AI Data Center Energy Demand Reshaping Power Infrastructure · history

Version 5

2026-05-25 11:11 UTC · 239 items

What

AI electricity demand is reshaping US power infrastructure across three simultaneous fronts: a proposed $67 billion NextEra–Dominion utility merger [13][14] whose regulatory fate rests primarily with Virginia's State Corporation Commission [17][18]; a parallel race toward grid-independent alternatives including onsite natural gas (now the default planning assumption for new training clusters [30]), fuel cells [34][37], and nuclear power [38][39]; and a widening structural mismatch between AI compute demand growing at 30–40% annually and US grid generation growing at roughly 2–3% [6], with PJM wholesale power prices already up 76% year-over-year in Q1 and capacity prices reported up by a factor of 10 [7][9] as evidence the strain has moved from forecast to market reality. Physical bottlenecks are sharpening: transformer lead times have reached 128 weeks or more [26][27], PJM and MISO interconnect queues were never designed for this scale of demand [5], and Denmark's grid operator has imposed a formal three-month moratorium on new connections after total requests reportedly reached 60 GW [41][42].

Why it matters

The gap between AI power demand growth and grid expansion capacity is no longer theoretical — it is appearing in both wholesale and capacity market prices, forcing hyperscalers to bypass the grid entirely, and triggering regulatory responses at federal, state, and now international levels. Who controls the infrastructure and who bears the cost will shape not just AI competition but US electricity affordability for years to come.

Open questions

  • Will Virginia's SCC impose conditions meaningful enough to protect consumers given NextEra's documented history in regulated markets — and does the 76% PJM wholesale price surge [7] and factor-of-10 capacity price increase [9] strengthen or complicate the case for merger-driven consolidation?

  • Can FERC's June 2026 large-load interconnection rulemaking [21] and the PJM–Google–Tapestry AI planning partnership [22][24] clear the backlog fast enough to slow the retreat to onsite gas generation [30], given that transformer lead times have reached 128 weeks and represent a hard physical constraint [26]?

  • Does onsite natural gas becoming the default planning assumption for AI training clusters [30] represent a durable structural decoupling from the grid — and what are the climate and regulatory implications if it does?

  • Can India, Denmark, Kenya, and other markets with acute grid constraints [57][41][45] sustain AI data center investment, or will chronic power shortfalls redirect global capital to better-powered markets?

Narrative

The United States power sector is undergoing structural transformation driven by AI compute. The US Energy Information Administration has forecast that domestic power demand will reach record highs in 2026 and 2027, with AI data centers identified as the primary driver [1]. Those facilities already consume approximately 4% of US electricity, a share analysts project will grow substantially [2][3][4]. SemiAnalysis has quantified the trajectory with unusual precision: US AI power demand has grown from roughly 3 GW in 2023 to a projected 28 GW by the end of 2026, with AI compute demand expanding at 30–40% annually while US grid generation grows at only about 2–3% [5][6]. Wholesale power prices on PJM — the US's largest electricity grid — surged 76% year-over-year in Q1 as AI-driven data center demand intensified [7][8], and IEEFA separately reports that projected data center growth has driven PJM capacity prices up by a factor of 10 [9], providing converging signals that grid strain has moved from projection to market reality. Analysts estimate that $800 billion in announced AI capex implies 50–70 GW of new power demand [10]. NVIDIA CEO Jensen Huang has projected that compute will eventually require 1,000 times more energy than currently produced [11]; Microsoft CEO Satya Nadella has crystallized the competitive dimension as 'Tokens per Dollar per Watt' [12].

The structural mismatch is being addressed through three competing strategies: grid consolidation, regulatory reform, and grid bypass. On the consolidation side, the proposed $67 billion NextEra–Dominion merger — the largest in US utility history, announced in May 2026 and explicitly motivated by data center electricity demand concentrated in northern Virginia [13][14][15] — would create the world's largest regulated electric utility by market value with a large-load pipeline exceeding 130 GW [16]. Legal analysts identify Virginia's State Corporation Commission as the single most consequential regulatory venue, with broader authority over Dominion's home territory than federal or other state regulators [17][18]. Opposition has escalated into antitrust territory: the American Economic Liberties Institute argues the deal would create a 'mega utility monopoly that makes families pay for the AI boom' [19], and Consumer Reports has documented how data center electricity costs flow through utility rate structures to residential bills [20]. On the reform side, FERC has committed to acting on a large-load interconnection rulemaking by June 2026 [21], and PJM has partnered with Google and the startup Tapestry to deploy AI itself to accelerate grid interconnection planning [22][23][24][25]. Physical bottlenecks constrain both paths: transformer supply lead times have reached 128 weeks — approximately two and a half years — with some sources citing ranges extending to four years [26][27][28][29], creating a hard constraint that capital alone cannot overcome.

The third strategy — grid bypass — has quietly become the industry default. SemiAnalysis reports that onsite natural gas generation has shifted from a fringe option to the default planning assumption for the next wave of US AI training clusters, driven by the inability to obtain grid interconnection on the timelines hyperscalers require [30][31][32][33]. Bloom Energy, which manufactures solid oxide fuel cells that generate electricity from natural gas without combustion, has secured a $7.65 billion data center contract pipeline [34][35], a 2.8 GW Oracle partnership [36], and an additional $2.65 billion deal with AEP for 1 GW of fuel cell capacity [37]. Small modular reactors have also entered the conversation: hyperscalers are actively pursuing nuclear power purchase agreements [38], and approximately 22 GW of SMR development for data centers has been reported [39], though skeptical analysts contend deployment timelines are too long and uncertain to address near-term demand [40].

AI's energy demand is straining power infrastructure internationally as well. Denmark's grid operator Energinet has imposed a formal three-month moratorium on new grid connection requests after total pending applications reportedly reached 60 GW — a figure that dwarfs the country's actual generating capacity [41][42][43][44]. Microsoft's planned billion-dollar data center in Kenya has sparked local blackout fears as the grid confronts the projected electricity draw [45][46]. India is targeting 10 GW of data center capacity by 2030 but confronts serious bottlenecks in power availability and grid infrastructure [47][48]. The environmental dimensions of AI's energy footprint remain contested: proponents of closed-loop cooling systems argue they substantially reduce freshwater consumption [49][50], while academic lifecycle analyses find meaningful trade-offs that vary by design and energy source [51][52][53]. Community opposition to data center development — citing power draw, noise, and light pollution alongside rate impacts — continues to intensify in affected localities [54][55][56].

Timeline

  • 2026-04-27: EIA forecasts US power demand will hit record highs in 2026–2027, driven by AI and data centers [1]
  • 2026-05-04: Denmark's grid operator Energinet imposes a formal three-month moratorium on new grid connection requests after total pending applications reportedly reach 60 GW [41][42][43][44]
  • 2026-05-11: US transformer market supply constraints reported with lead times reaching 128 weeks; pv magazine and industry sources flag the shortage as a hard bottleneck on grid expansion [27][26][28][29]
  • 2026-05-17: Satya Nadella's 'Tokens per Dollar per Watt' framing amplified as the defining competitive metric of the AI era; transformer supply lead times confirmed at 2–4 years [12][71][72]
  • 2026-05-18: Microsoft's Kenya AI data center sparks blackout fears as local grid struggles with projected electricity demand [45][46]
  • 2026-05-19: NextEra and Dominion announce $67 billion merger — the largest in US utility history — explicitly driven by data center electricity demand in northern Virginia; PJM unveils plan to tackle AI-driven power demand surge [13][14][15][73][58][74]
  • 2026-05-20: Constellation and Vistra stocks rally as PJM accelerates data-center deals; analyst notes NextEra is targeting Dominion's PJM interconnect position [75][76]
  • 2026-05-21: Greg Brockman challenges the 'running tap' narrative on AI water consumption, arguing closed-loop cooling systems recirculate rather than continuously draw fresh water [49]
  • 2026-05-22: Bloom Energy's $7.65 billion data center fuel cell contract pipeline reported; Jensen Huang's 1,000x energy projection cited as the demand thesis [11][34][35]
  • 2026-05-23: SemiAnalysis quantifies US AI power demand trajectory at 3 GW (2023) to 28 GW (end 2026) and identifies onsite gas as the new default for training cluster planning; Barron's and WSJ detail the multi-agency regulatory gauntlet for the NextEra-Dominion deal; Energy and Policy Institute publishes critique of NextEra's rate increase history [5][30][61][59][62]
  • 2026-05-24: American Economic Liberties Institute frames NextEra-Dominion as a 'mega utility monopoly'; legal analysts identify Virginia's SCC as decisive venue; Consumer Reports documents household electricity bill impacts; PJM wholesale power prices reported up 76% YoY in Q1; IEEFA separately reports PJM capacity prices up by factor of 10 [19][17][18][20][7][9]
  • 2026 ongoing: FERC committed to acting on large-load interconnection rulemaking by June 2026; PJM partners with Google and Tapestry to use AI for interconnection planning; Bloom Energy adds $2.65B AEP deal for 1 GW SOFC capacity; SMR development pipeline for data centers reported at approximately 22 GW [21][22][23][24][25][37][39]

Perspectives

Satya Nadella (Microsoft CEO)

Energy efficiency is a primary competitive dimension; 'Tokens per Dollar per Watt' is the key metric; infrastructure investment is the dominant strategic priority of the AI era

Evolution: Consistent

[12]

Jensen Huang (NVIDIA CEO)

Compute will require 1,000x more energy than currently produced; even that estimate may be too conservative

Evolution: Consistent

[11]

SemiAnalysis

US AI power demand has grown from ~3 GW in 2023 to a projected 28 GW by end 2026, a step-function that PJM and MISO interconnect queues were never designed to absorb; onsite natural gas generation has quietly become the default planning assumption for the next wave of US training clusters as hyperscalers bypass grid interconnection backlogs

Evolution: Consistent since introduction; provides the most precise demand-side framing in the thread

[5][30]

NextEra Energy / Dominion Energy

The $67 billion merger is the appropriate response to surging data center electricity demand; the combined entity would carry a 130+ GW large-load pipeline and serve the public interest

Evolution: Consistent

[13][16][58]

American Economic Liberties Institute

The NextEra-Dominion merger would create a 'mega utility monopoly' that forces families to subsidize the AI boom — an antitrust threat, not merely a consumer rate concern

Evolution: Consistent

[19]

Consumer advocates and Energy and Policy Institute

The NextEra-Dominion merger would harm consumers and undermine environmental accountability; NextEra's documented history of political lobbying and rate increases in regulated markets is a specific red flag; the 76% YoY PJM wholesale price surge and factor-of-10 capacity price increase provide market-level corroboration of predicted cost transmission to ratepayers

Evolution: Consistent; IEEFA's factor-of-10 capacity price data adds a forward-looking capacity market dimension to the rate-harm argument alongside the previously reported wholesale price surge

[13][59][60][19][20][7][9]

Legal analysts (Whiteford / Williams Mullen)

The NextEra-Dominion merger will be decided primarily in Virginia; Virginia's SCC is the single most consequential regulatory venue, with broader authority over Dominion's home territory than federal or other state regulators

Evolution: Consistent

[17][18]

Barron's / Wall Street Journal

The regulatory path for the NextEra-Dominion deal is long and uncertain, requiring FERC, multiple state commissions, and potentially congressional approval — the deal is far from done

Evolution: Consistent

[61][62]

Consumer Reports

AI data center development has measurable impacts on household electricity bills and water use; these costs are borne by ordinary ratepayers

Evolution: Consistent

[20]

Greg Brockman (OpenAI) / Oracle

Closed-loop cooling systems substantially reduce AI's freshwater footprint by recirculating stored water; Oracle's operational adoption corroborates the technical claim

Evolution: Consistent

[49][50]

Bloom Energy

Solid oxide fuel cells offer grid-independent, fast-deployment power for AI data centers; the combined $7.65B contract pipeline including a 2.8 GW Oracle partnership and a $2.65B AEP deal for 1 GW demonstrates commercial viability at scale

Evolution: Consistent

[34][35][36][63][64][65][37]

Nuclear optimists (Introl / hyperscaler procurement advocates)

Nuclear power including SMRs represents a major opportunity for AI data center power, with approximately 22 GW of SMR development reportedly underway; hyperscaler procurement interest is creating a genuine market that will reshape AI energy supply

Evolution: Consistent

[38][66][39]

Nuclear skeptics (Tony Grayson / independent analysts)

SMR deployment timelines are too long and uncertain to match the hype; nuclear reality does not match the data center promise in the near term

Evolution: Consistent

[40]

US Energy Information Administration (EIA)

US power demand will reach record highs in 2026–2027, driven by AI data centers — the most authoritative official forecast anchoring the demand side of the debate

Evolution: Consistent

[1]

FERC

The federal regulator is committed to acting on a large-load interconnection rulemaking by June 2026 and has directed PJM to adopt new rules embracing innovation; FERC is also the required federal approval body for the NextEra-Dominion merger

Evolution: Consistent

[21][67][68]

PJM / Google / Tapestry

Grid interconnection delays can be addressed using AI-assisted planning tools; PJM's partnership with Google and Tapestry represents a technical approach to clearing the interconnection backlog rather than waiting for physical infrastructure expansion

Evolution: Consistent

[22][23][24][69][25]

Denmark / Energinet

The scale of pending data center grid connection requests — reportedly 60 GW against a far smaller national grid — has forced a formal three-month moratorium; Denmark's case illustrates that even grids with strong renewable buildouts can be overwhelmed by concentrated AI demand

Evolution: Confirmed as formal moratorium; prior synthesis reported the pause, but the three-month formal moratorium structure is now documented

[41][42][43][44][57]

India data center market / Analytics India Magazine

India is targeting 10 GW of data center capacity by 2030 but faces serious bottlenecks in power and infrastructure that threaten that timeline

Evolution: Consistent

[47][70][48]

Community and environmental opposition (local residents, NRDC)

AI data center expansion imposes unacceptable costs on communities through power draw, noise, and light pollution; grid stress does not need to be resolved through utility consolidation that raises rates

Evolution: Consistent

[54][55][56]

IEEFA (Institute for Energy Economics and Financial Analysis)

Projected data center growth has driven PJM capacity prices up by a factor of 10, providing a forward-looking capacity market signal that complements the reported 76% wholesale price surge and underscores the systemic scale of AI's impact on grid pricing

Evolution: First appearance as a distinct voice; adds a capacity market dimension — distinct from spot wholesale prices — to the market-impact evidence base

[9]

Tensions

  • Tech leaders and investors (Nadella, Huang, Bloom Energy) frame AI energy demand as a strategic imperative requiring massive infrastructure build-out, while consumer advocates, the American Economic Liberties Institute, Consumer Reports, IEEFA, and community opponents frame the same demand surge as enabling harmful utility consolidation and forcing ordinary families to subsidize the AI boom — a tension now backed by both a 76% YoY PJM wholesale price surge and a factor-of-10 capacity price increase [12][13][11][59][19][20][54][60][7][9]
  • NextEra and Dominion argue their merger is the necessary and appropriate response to AI data center power demand; legal analysts, the American Economic Liberties Institute, the Energy and Policy Institute, Barron's, and the WSJ contend that Virginia's SCC is the decisive venue where NextEra's track record in regulated markets should give regulators and consumers serious pause [13][16][61][59][62][17][18][19]
  • SemiAnalysis and onsite gas proponents argue that grid bypass via onsite natural gas is already the de facto default for new AI training clusters given interconnect backlogs, while grid consolidation advocates and FERC reformers argue the grid can and must absorb AI demand through expansion and regulatory innovation rather than circumvention [30][13][16][21][22][23][24][31][32][33]
  • Bloom Energy and off-grid fuel cell advocates position distributed generation as the answer to grid interconnection backlogs — implicitly challenging utility consolidation as the primary structural response to AI power demand [34][35][36][63][13][16][64][65][37]
  • Nuclear optimists project a multi-GW SMR-powered data center revolution with approximately 22 GW in development, while nuclear skeptics argue that deployment timelines are too long and uncertain to address near-term AI power demand [38][40][66][39]
  • Greg Brockman and Oracle argue that closed-loop cooling systems substantially reduce AI's freshwater footprint, while academic lifecycle analyses and water-treatment specialists note that closed-loop systems introduce chemical and operational costs of their own — leaving the net environmental impact genuinely contested [49][50][51][52][53]

Sources

  1. [1] US power demand to reach record highs in 2026–2027 driven by AI and data centers — reactive:ai-energy-infrastructure (2026-04-27)
  2. [2] Energy demand from AI – Energy and AI – Analysis - IEA — reactive:ai-energy-infrastructure
  3. [3] Electricity Demand and Grid Impacts of AI Data Centers - arXiv — reactive:ai-energy-infrastructure
  4. [4] @iatnon @zerohedge **Short-term impact:** Massive strain on the US grid. AI data centers are already ~4% of US electrici... — reactive:ai-energy-infrastructure (2026-05-21)
  5. [5] The basic shape of the problem is that US AI power demand has run from roughly 3 GW in 2023 to a path of about 28 GW by … — SemiAnalysis Twitter (2026-05-23)
  6. [6] @NoLimitGains The grid math is brutal. AI compute demand is growing 30–40% annually while US grid generation is growing ... — reactive:ai-power-grid-crisis (2026-05-18)
  7. [7] BREAKING: Wholesale power prices on PJM, America’s largest electricity grid, surged 76% YoY in Q1 as AI-driven data cent... — reactive:ai-power-grid-crisis (2026-05-24)
  8. [8] PJM Power Prices Up 76% From Data Centers: CRE Guide — reactive:ai-energy-infrastructure
  9. [9] Projected data center growth spurs PJM capacity prices by factor of 10 — reactive:ai-energy-infrastructure
  10. [10] @DariusDale42 The power demand angle gets buried in most AI inflation analyses. $800B in AI capex implies 50-70 GW of ne... — reactive:ai-power-grid-crisis (2026-05-24)
  11. [11] Bloom Energy is one of the most interesting AI infrastructure plays most people still haven't fully priced in (Save this… — Milk Road AI Twitter (2026-05-22)
  12. [12] Satya Nadella's energy is something here. 🔥 — Rohan Paul Twitter (2026-05-17)
  13. [13] Electrical utility megamerger is all about the data centers — Ars Technica AI (2026-05-19)
  14. [14] NextEra to Buy Dominion in $66.8B Deal to Power AI Data Centers - BIC Magazine — reactive:ai-energy-infrastructure
  15. [15] NextEra bets $66.8B on AI power boom with Dominion Energy acquisition | Fox Business — reactive:ai-energy-infrastructure
  16. [16] Combined NextEra-Dominion would have 130-GW large-load pipeline — reactive:ai-energy-infrastructure
  17. [17] Client Alert: The NextEra-Dominion Merger Will Be Decided in Virginia | Whiteford - JDSupra — reactive:ai-energy-infrastructure
  18. [18] NextEra Energy’s Proposed Acquisition of Dominion Energy – Virginia State Corporation Commission Regulatory Approval Process | Williams Mullen — reactive:ai-energy-infrastructure
  19. [19] NextEra-Dominion Merger Would Create a Mega Utility Monopoly That Makes Families Pay for the AI Boom - NextEra-Dominion Merger Would Create a Mega Utility Monopoly That Makes Families Pay for the AI Boom — reactive:ai-energy-infrastructure
  20. [20] AI Data Centers: Big Tech's Impact on Electric Bills, Water, and More — reactive:ai-energy-infrastructure
  21. [21] FERC to Act on Large Load Interconnection Docket by June 2026 | Federal Energy Regulatory Commission — reactive:ai-power-grid-crisis
  22. [22] PJM, Google & Tapestry Join Forces To Apply AI To Enhance Regional Planning, Generation Interconnection | PJM Inside Lines — reactive:ai-power-grid-crisis
  23. [23] Tapestry is using AI to help PJM clear its interconnection backlog | Latitude Media — reactive:ai-power-grid-crisis
  24. [24] PJM, Google partner to speed grid interconnection using AI — reactive:ai-power-grid-crisis
  25. [25] PJM Taps Google and Tapestry to Use AI for Grid Interconnection Planning — reactive:ai-power-grid-crisis
  26. [26] Power Transformer Lead Times Hit 128 Weeks in 2026 — reactive:ai-energy-infrastructure
  27. [27] U.S. transformer market faces severe supply constraints as lead times extend to four years – pv magazine USA — reactive:ai-energy-infrastructure
  28. [28] Transformer shortage strains US grid – Transformer Magazine — reactive:ai-energy-infrastructure
  29. [29] 24+ Month Lead Times: New Normal for Transformer Suppliers — reactive:ai-energy-infrastructure
  30. [30] One of the threads we kept pulling on in our recent piece on how AI labs are solving the power crisis is that onsite gas… — SemiAnalysis Twitter (2026-05-23)
  31. [31] Gas-to-Power Boom: AI Drives 2026 On-Site Energy Shift — reactive:ai-power-grid-crisis
  32. [32] The rise of on-site generation to power AI | Latitude Media — reactive:ai-power-grid-crisis
  33. [33] How AI Labs Are Solving the Power Crisis: The Onsite Gas Deep Dive — reactive:ai-power-grid-crisis
  34. [34] Bloom Energy Fuel Cell 2026, $7.65B Data Center Deals — reactive:ai-energy-infrastructure
  35. [35] Fuel Cells: AI Data Center Power's $7.65B Dark Horse | Introl Blog — reactive:ai-energy-infrastructure
  36. [36] Bloom Energy and Oracle Expand Strategic Partnership to Deploy ... — reactive:ai-energy-infrastructure
  37. [37] Bloom Energy secures $2.65B AEP deal for 1GW SOFC capacity — reactive:ai-energy-infrastructure
  38. [38] Nuclear Power for Data Centers: What the Hyperscaler Procurement Rush Means for Developers | Insights | Build — reactive:ai-energy-infrastructure
  39. [39] SMR Nuclear for Data Centers Accelerates | Introl Blog — reactive:ai-energy-infrastructure
  40. [40] The SMR Data Center Promise: Why Nuclear Reality Doesn't Match ... — reactive:ai-energy-infrastructure
  41. [41] Danish grid operator introduces three-month moratorium for new grid connections - DCD — reactive:ai-energy-infrastructure
  42. [42] Denmark presses pause on new data center grid connections as ... — reactive:ai-energy-infrastructure
  43. [43] Denmark faces data center reckoning amid power grid strains — reactive:ai-energy-infrastructure
  44. [44] Denmark pauses grid connections as AI data centres overwhelm the cleanest power grid in Europe — reactive:ai-energy-infrastructure
  45. [45] Microsoft’s billion-dollar Kenya AI data center sparks blackout fears as power grid struggles with massive electricity d... — reactive:ai-power-grid-crisis (2026-05-18)
  46. [46] Microsoft’s billion-dollar Kenya AI data center sparks blackout fears as power grid struggles with massive electricity d... — reactive:ai-power-grid-crisis (2026-05-18)
  47. [47] 10 GW of data centre capacity by 2030. That's the scale India is ... — reactive:ai-energy-infrastructure
  48. [48] India's Data Centre Growth Faces Bottlenecks in Power and Infrastructure | AIM posted on the topic | LinkedIn — reactive:ai-energy-infrastructure
  49. [49] Greg Brockman explains how the public story about AI data center water use is partly wrong. — Rohan Paul Twitter (2026-05-21)
  50. [50] Filled once. Recirculated continuously. Our closed-loop cooling ... — reactive:ai-energy-infrastructure
  51. [51] Liquid versus Air: Life Cycle Carbon of Cooling Down AI Data Centers — reactive:ai-energy-infrastructure
  52. [52] Closed-Loop Cooling: Water Saver or Chemical Time Bomb? | KETOS — reactive:ai-energy-infrastructure
  53. [53] The carbon and water footprints of data centers and what this could ... — reactive:ai-energy-infrastructure
  54. [54] We don't want AI infrastructure if it means unchecked water and power consumption, noise pollution, light pollution, los... — reactive:ai-energy-infrastructure (2026-05-18)
  55. [55] The AI Boom Is Stressing the Grid—but It Doesn't Have to Be This Way — reactive:ai-energy-infrastructure
  56. [56] Resist #AI #DataCenters! — reactive:ai-energy-infrastructure (2026-05-22)
  57. [57] AI/POWER: Denmark paused new data center grid connections after total requests reportedly reached 60 GW. — reactive:ai-power-grid-crisis (2026-05-17)
  58. [58] Dominion Energy agrees to merger with NextEra Energy | D SEC Filing - Form 425 — reactive:ai-energy-infrastructure
  59. [59] NextEra’s acquisition of Dominion would bring history of political control, rate increases to Virginia, Carolinas - Energy and Policy Institute — reactive:ai-energy-infrastructure
  60. [60] NextEra-Dominion Merger Would Create a Mega Utility Monopoly That Makes Families Pay for the AI Boom https://t.co/biM2Ha... — reactive:ai-energy-infrastructure (2026-05-20)
  61. [61] The Long Regulatory Road Ahead for the NextEra-Dominion Deal — reactive:ai-energy-infrastructure
  62. [62] The Biggest Challenge of a Utility Megadeal: Regulators - WSJ — reactive:ai-energy-infrastructure
  63. [63] AEP, Bloom Energy 1-GW fuel cell deal to power data centers would ... — reactive:ai-energy-infrastructure
  64. [64] [PDF] 2026 Data Center Power Report - Bloom Energy — reactive:jensen-huang-nvidia-thesis
  65. [65] Powering AI: Future-Proofing Data Centers - Bloom Energy — reactive:ai-energy-infrastructure
  66. [66] SMRs Power AI: $10B Nuclear Data Center Revolution | Introl Blog — reactive:ai-energy-infrastructure
  67. [67] FERC Targets Data Center Interconnection Delays — reactive:ai-power-grid-crisis
  68. [68] FERC Directs Nation's Largest Grid Operator to Create New Rules to ... — reactive:ai-power-grid-crisis
  69. [69] Our investment in AI-powered solutions for the electric grid — reactive:ai-power-grid-crisis
  70. [70] India Data Center Power Market Size, Trends Report 2026-2034 — reactive:ai-energy-infrastructure
  71. [71] Transformer supply bottleneck threatens power system stability as ... — reactive:ai-power-grid-crisis
  72. [72] The One Device Throttling the World's Electrified Future — reactive:ai-power-grid-crisis
  73. [73] Dominion Energy, NextEra $66.8B merger is powered by AI expansion https://t.co/xPqH0ds884 — reactive:ai-energy-infrastructure (2026-05-19)
  74. [74] PJM unveils plan to tackle AI-driven power demand surge | Reuters — reactive:ai-power-grid-crisis
  75. [75] 🚨 CONSTELLATION’S AND VISTRA’S STOCKS RALLY AS POWER-GRID OPERATOR SPEEDS UP DATA-CENTER DEALS — reactive:ai-power-grid-crisis (2026-05-20)
  76. [76] @CaseyVSilver AI compute meets utility consolidation, playing out in real time. NextEra wants Dominion's PJM interconnec... — reactive:ai-power-grid-crisis (2026-05-18)