The Grid Report

AI Energy Infrastructure AI Automation Markets Policy Tracker

Daily intelligence on AI infrastructure

The Grid Report

Power, grid, and load growth

Energy and grid news shaping AI infrastructure deployment

Electricity, substations, generation, and the physical constraints shaping what AI can scale into.

Latest stories

All Energy coverage

32 stories

Editorial graphic showing a PJM transmission line fault, a large AI data center dropping load to backup power, grid-frequency and voltage disturbance indicators, and a new contract layer for ride-through, telemetry, reactive support, and control settings

Energy GridJuly 24, 20264 min read

PJM’s Data-Center Load-Drop Risk Turns AI Power Quality Into a Grid-Contract Problem

This July 24, 2026 story clears the publish bar because it is not another broad article about AI power demand. The stronger signal is that grid operators are being forced to care about how hyperscale AI loads behave during faults, not just how many megawatts they want. In practice, that pushes data-center siting toward a new contract stack around ride-through, dynamic models, reactive support, monitoring, and backup-power coordination.

By Nawaz Lalani

Power-quality contract

Editorial graphic showing GE Vernova gas turbine slots, data center power demand, transformer and switchgear backlog, and a sold-through timeline stretching into 2031

Energy GridJuly 23, 20264 min read

GE Vernova’s Q2 Turns AI Power Into a Turbine-Slot and Grid-Equipment Queue

GE Vernova’s July 22, 2026 results clear the publish bar because this is not just another industrial earnings beat. The stronger signal is that AI power buildout is hardening into a manufacturing queue for gas turbines, transformers, and switchgear, which means power readiness now depends as much on equipment-slot timing as on land, interconnection, and capital.

By Nawaz Lalani

Turbine queue

Editorial graphic showing OpenAI Project Camellia in Georgia connected to 3.2 gigawatts of phased power, one gigawatt of demand reduction, ratepayer protections, closed-loop cooling, and an independent annual audit

Energy GridJuly 22, 20264 min read

OpenAI’s Project Camellia Turns 3.2 GW of AI Load Into a Community Operating Contract

OpenAI’s July 22, 2026 plan for Project Camellia clears the publish bar because it does more than announce another giant campus. The stronger grid signal is the operating package around the 3.2-gigawatt load: full infrastructure cost recovery, up to 1 gigawatt of peak demand reduction, closed-loop cooling, community benefits, and an independent annual audit.

By Nawaz Lalani

Grid contract

Editorial graphic showing Quebec hydropower flowing through the CHPE transmission line into New York City, with import volumes, reliability support, and lower-cost downstate power highlighted across the route

Energy GridJuly 20, 20264 min read

New York’s Canada Import Spike Turns CHPE Into a Downstate Reliability-and-Cost Hedge

EIA’s July 20, 2026 trade data clears the bar because it is not just another clean-energy progress note. The stronger grid signal is that the newly energized Champlain Hudson Power Express is already changing New York City’s supply stack: transmission imports from Quebec are becoming a live reliability and cost hedge for downstate load, not merely a climate-policy promise.

By Nawaz Lalani

Import hedge

Editorial graphic showing New York pausing new hyperscale data centers while directing large loads to fund grid upgrades, clean generation, and community investment

Energy GridJuly 19, 20264 min read

New York’s Data-Center Moratorium Turns AI Buildout Into a Grid-and-Ratepayer Test

New York’s July 14 moratorium clears the bar because it is not just another anti-data-center backlash story. The sharper signal is that a major state is explicitly testing whether hyperscale AI load should fund its own grid upgrades, clean supply, and community offsets before the next buildout wave is allowed to proceed.

By Nawaz Lalani

Large-load test

Editorial graphic showing Google’s Steel River Energy Center linking solar, storage, local grid support, Arkansas steel, and community affordability funding into one AI power stack

Energy GridJuly 18, 20264 min read

Google’s Steel River Project Turns AI Power Procurement Into a Community Grid Product

Google’s July 14 Steel River announcement clears the bar because it is not just another giant clean-power purchase. The stronger signal is that hyperscaler energy buying is being packaged as a regional grid-and-community product: new solar and storage, domestic supply-chain framing, local affordability funding, and political durability all attached to the same AI-era power stack.

By Nawaz Lalani

Regional power stack

Editorial graphic showing PJM record July heat load, demand response, forced outages, DOE emergency orders, and data center backup generation as one operating stack

Energy GridJuly 11, 20265 min read

PJM's Record July Heat Load Turns Demand Response Into a Capacity-Planning Signal

PJM's July 10 after-action update clears the bar because it converts the early-July heat wave from an alert story into a planning signal: the system served a preliminary 168,158 MW record only after demand response, emergency procedures, every available generator, and DOE backup-generation authority were lined up together.

By Nawaz Lalani

Peak load

High-voltage transmission corridor representing DOE’s July 2026 view that AI load growth is exposing an interregional transmission build problem, not only local utility constraints

Energy GridJuly 9, 20266 min read

DOE’s July 2026 Transmission Study Turns AI Power Into an Interregional Grid-Build Problem

DOE’s July 9, 2026 transmission-needs release clears the bar because it does more than restate that load is growing. The stronger signal is that data-center and AI demand are now exposing a national transmission problem: the highest-value fixes sit across market seams and interregional transfer paths, not only inside local utility footprints.

By Nawaz Lalani

Grid seams

Editorial graphic showing Meta’s Alberta data center linked to a 932 megawatt gas plant, Alberta’s bring-your-own-power model, and grid protection rules

Energy GridJuly 8, 20266 min read

Meta’s Alberta Campus Turns Off-Grid AI Power Into a Jurisdiction Product

Meta’s July 8, 2026 Alberta announcement clears the bar because it is not just another hyperscale campus reveal. The stronger angle is that Alberta is trying to sell AI developers a different product: bring your own generation, avoid stressing the shared grid, and get a faster path to scale inside an industrial zone built for heavy infrastructure.

By Nawaz Lalani

Bring your own power

High-voltage transmission towers under summer conditions, illustrating grid stress and real-time power delivery during heat-wave operations

Energy GridJuly 8, 20266 min read

PJM’s Heat-Wave Warning Put Data Center Curtailment Into Real Grid Operations

PJM’s July 2 and July 3 operating updates matter because they moved large-load flexibility out of planning decks and into live grid operations: the operator prepared utilities for the possibility of curtailing data centers and shifting them to backup generation while demand pushed near record levels.

By Nawaz Lalani

Grid operations

Editorial graphic showing server racks, a one-third commercial power meter, and a flat 24-7 load curve based on EIA data center demand forecasts

Energy GridJuly 3, 20265 min read

EIA’s Data Center Server Forecast Says AI Could Reach One-Third of Commercial Power Demand

EIA’s May 19 data-center server forecast clears the publish bar because it is not another loose AI-power estimate. The stronger angle is that EIA is now modeling server demand as a distinct commercial end use with a flat load shape, rising cooling needs, and a scenario where server electricity alone reaches 22% to 33% of all U.S. commercial-building electricity use by 2050.

By Nawaz Lalani

Load-shape warning

Residential electric meter and utility service equipment mounted on a building exterior

Energy GridJuly 2, 20265 min read

New Jersey’s Data-Center Tariff Bill Turns AI Load Into a Rate-Segregation Test

New Jersey’s June 30 large-load bill clears the bar because it is not another vague ratepayer-warning story. The stronger angle is that the state is trying to force AI-scale load into its own tariff logic before transmission, distribution, capacity, and stranded-cost risk spread across ordinary customers.

By Nawaz Lalani

Rate segregation

Editorial graphic showing PJM moving new large loads into a connect-and-manage framework with curtailment terms, matched supply, and backstop procurement

Energy GridJuly 2, 20265 min read

PJM’s Connect-and-Manage Push Turns Large-Load Growth Into a Curtailable Grid-Service Model

PJM’s July 1 stakeholder update clears the bar because it is not just another process note about data-center demand. The stronger angle is that PJM is moving toward a service model in which some new large loads can connect earlier only if they accept curtailment risk, matched supply expectations, and a more explicit reliability bargain.

By Nawaz Lalani

Curtailable load

Editorial graphic showing Google AI infrastructure load growth rising faster than grid decarbonization, with clean energy procurement and power bottlenecks in tension

Energy GridJuly 1, 20265 min read

Google’s 37% Electricity Jump Turns AI Growth Into a Grid Decarbonization Race

Google’s 2026 Environmental Report clears the bar because it is one of the cleanest primary-source disclosures yet on the AI power problem. The stronger angle is not that Google bought more clean energy. It is that the company’s largest electricity-demand increase in history arrived while Google says AI infrastructure is growing faster than the grid can decarbonize.

By Nawaz Lalani

AI load growth

Editorial graphic showing DOE loan financing flowing into AP1000 reactor components, long-lead manufacturing, and multi-gigawatt power delivery

Energy GridJune 24, 20265 min read

DOE’s $17.5 Billion Nuclear Supply-Chain Loans Turn AI Power Into a Long-Lead Manufacturing Story

DOE’s June 23 loan commitment clears the bar because it does not just add another pro-nuclear headline. The stronger angle is that the department is financing the slowest manufacturing layer in firm power: long-lead reactor components that can decide whether large-load growth gets real capacity on time.

By Nawaz Lalani

Firm power

Editorial graphic showing PJM emergency procedures moving from all available generation through large-load backup generation to voltage reduction and manual load dump

Energy GridJune 24, 20265 min read

PJM’s Manual 13 Update Turns Large-Load Backup Generation Into a Grid Emergency Tool

PJM’s June 24 emergency-procedures change clears the publish bar because it does more than extend another reliability warning. The stronger angle is that the grid operator is formalizing a pre-blackout step in which some data centers and other large loads can be directed to move onto backup generation before voltage reduction or manual load shed begins.

By Nawaz Lalani

Emergency stack

Editorial graphic showing Google’s Alabama data center expansion, self-funded power costs, and community energy offsets

Energy GridJune 23, 20264 min read

Google’s Alabama Expansion Turns AI Campus Growth Into a Ratepayer-Protection Bargain

Google’s June 15 Alabama expansion clears the bar because it does not present another generic data-center investment. The stronger angle is that Google is explicitly trying to package campus growth with a ratepayer-protection argument by funding 100% of its own power and infrastructure costs and pairing the buildout with a local energy-efficiency fund.

By Nawaz Lalani

Ratepayer shield

Editorial graphic showing DOE federal sites, transmission and generation assets, nuclear permitting pathways, and AI data center campuses moving through a faster speed-to-power funnel

Energy GridJune 23, 20264 min read

DOE’s AI-on-DOE-Lands RFI Turns Federal Sites Into a Speed-to-Power Product

DOE’s AI-on-DOE-lands request for information clears the bar because it does more than advertise surplus federal acreage. The stronger angle is that Washington is trying to convert federal sites with existing energy infrastructure into a faster path for AI load, new generation, and data-center construction.

By Nawaz Lalani

Federal siting play

Editorial graphic showing TVA’s 2026 integrated resource plan, rising data center demand, public-comment timing, and gas optionality inside a broader Southeast power portfolio

Energy GridJune 23, 20264 min read

TVA’s 2026 IRP Turns Southeast AI Power Into a Gas-Optionality Story

TVA’s preliminary final 2026 integrated resource plan clears the bar because the federal utility explicitly says actual and forecasted demand is rising primarily from population and data center growth. The stronger angle is not that TVA has chosen one fuel. It is that one of the South’s most important power systems is preserving large dispatchable optionality while AI-era load remains uncertain in timing, size, and persistence.

By Nawaz Lalani

Southeast load signal

Editorial graphic showing ERCOT solar generation overtaking coal alongside AI load curves and timing constraints

Energy GridJune 23, 20264 min read

ERCOT’s Solar-Coal Crossover Turns Texas AI Power Into a Timing Problem

EIA’s May 13, 2026 ERCOT forecast clears the bar because it changes the shape of the Texas power story. The stronger angle is not that solar is winning a headline. It is that AI load in Texas is becoming less a volume problem and more a timing, storage, and firm-capacity problem.

By Nawaz Lalani

Timing over volume

Editorial graphic showing PJM queue reform, faster generation studies, and the next bottlenecks shifting into permitting, equipment, and project execution

Energy GridJune 22, 20265 min read

PJM’s New Interconnection Process Turns the AI Power Bottleneck Into a Project-Execution Story

PJM’s June 17, 2026 interconnection update clears the bar because it shifts the AI-power discussion from queue backlog rhetoric to a harder operating fact: even after process reform, the next constraint is whether generation projects can clear permitting, siting, and supply-chain delays fast enough to match data-center demand.

By Nawaz Lalani

Execution bottleneck

Editorial graphic showing Permian natural gas output growth, higher gas-oil ratios, and the fuel base behind Texas AI campus power

Energy GridJune 22, 20264 min read

EIA’s Permian Gas Data Turns Texas AI Power Into a Fuel-Supply Story

EIA’s June 18, 2026 Permian update matters because it shifts part of the Texas AI-power debate away from queue theory and toward fuel reality. If associated-gas output and pipeline takeaway keep growing, the practical bottleneck for new campuses may become how quickly developers can convert gas abundance into bankable, dispatchable power.

By Nawaz Lalani

Fuel before wires

Editorial graphic showing Texas large-load requests flowing into ERCOT Batch Zero, then splitting into base load, studied load, and excluded projects with self-supply and curtailment pathways

Energy GridJune 22, 20264 min read

Texas PUCT’s Batch Zero Approval Turns AI Power Into a Qualification-and-Curtailment Story

Texas’ June 18, 2026 approval of ERCOT’s Batch Zero framework matters because the state is no longer treating giant data-center loads like ordinary one-off requests. It is forcing them through a qualification system that decides who keeps prior standing, who gets partial capacity, who gets deferred, and which projects can move faster only by bringing their own power or agreeing to be curtailed.

By Nawaz Lalani

Texas queue reset

Editorial graphic showing Google-funded distributed energy resources flowing through a Voltus virtual power plant into PJM grid capacity that supports data center growth

Energy GridJune 21, 20264 min read

Google’s Voltus Deal Turns AI Load Into a PJM Capacity Product

Google and Voltus’s June 2, 2026 agreement matters because it reframes one piece of AI power growth: instead of treating new data-center demand only as a generation-and-transmission buildout problem, Google is buying distributed flexibility as a grid-capacity product.

By Nawaz Lalani

Flexible capacity

Transmission infrastructure representing NERC’s Level 3 alert that AI data centers are becoming grid reliability actors requiring better modeling, controls, and commissioning

Energy GridMay 14, 20267 min read

NERC’s Level 3 Alert Turns AI Data Centers Into a Grid Reliability Problem

NERC’s May 2026 computational-load alert is a quiet but important shift: AI data centers are no longer being treated only as big electricity customers. They are becoming reliability actors that utilities and grid operators need to model, instrument, commission, and control more carefully.

By Nawaz Lalani

Grid reliability

Utility grid infrastructure representing the core question of how much electricity AI and data centers are consuming in 2026 and where that demand is landing

Energy GridMay 13, 20267 min read

How Much Electricity Does AI Actually Use in 2026?

The honest answer is no longer “a lot” or “not that much.” AI electricity use is rising quickly, but the real story depends on the difference between training and inference, how much load lands in data centers, and how fast grids can absorb new demand.

By Nawaz Lalani

Energy explainer

Residential electricity meter and utility equipment mounted on the outside of a home

Energy GridMay 13, 20267 min read

Will Your Electricity Bill Go Up Because of AI?

The uncomfortable answer is yes in some places, but not in the simple way most people assume. Household bills are more likely to rise when utilities socialize new infrastructure costs, when capacity-market rules price in projected AI load early, or when local grid bottlenecks force expensive upgrades.

By Nawaz Lalani

Ratepayer explainer

Power infrastructure and generation assets representing why AI electricity demand is now best understood as a planning range rather than a single forecast number

Energy GridMay 10, 20267 min read

The AI Power Forecast Is Now a Planning Range, Not a Single Number

The strongest AI electricity story is no longer one scary demand estimate. It is the spread between scenarios. IEA says global data center electricity use could nearly double by 2030, while EIA expects U.S. power demand to keep rising as large computing facilities expand. The operating question is which projects become real load, where they land, and whether the grid can stage capacity fast enough.

By Nawaz Lalani

Weekly data feature

Transmission infrastructure representing FERC turning AI data-center power into a large-load rules, interconnection, and cost-allocation problem

Energy GridMay 10, 20266 min read

FERC’s Large-Load Clock Turns AI Power Into a Rules-and-Cost Allocation Story

AI data center demand is no longer only a utility forecast problem. FERC’s large-load interconnection work shows the next constraint is governance: who studies massive new loads, who pays for upgrades, how quickly they can connect, and what happens when speculative projects crowd the queue.

By Nawaz Lalani

Grid policy brief

Transmission towers and power lines crossing an open landscape

Energy GridMay 9, 20266 min read

Flexible AI Factories Are Becoming a Grid Product, Not Just a Data Center Design

NVIDIA, Emerald AI, and a group of major energy companies are making a more serious claim than it first appears. If AI campuses can flex power use, pair with co-located generation and storage, and connect faster, they stop looking like passive loads and start looking like a new kind of grid-facing infrastructure product.

By Nawaz Lalani

Energy analysis

Electric substation and transmission towers representing utility grid planning

Energy GridMay 7, 20266 min read

Utilities Need an AI Load Playbook, Not Just More Optimism

The AI power story is moving faster than many utility planning processes were built for. The hard part is no longer acknowledging demand growth. It is building a playbook for interconnection, load timing, reliability, and who gets prioritized when capacity tightens.

By Nawaz Lalani

Energy-grid analysis

Electric vehicle charging outdoors to illustrate the emissions, energy-system, and charging-infrastructure tradeoffs behind EV adoption

Energy GridApril 4, 20267 min read

How Much Carbon Do EVs Really Save Versus Gas Cars — and Is Buying One Feasible in a More Unstable World?

Electric vehicles usually beat gasoline cars on lifecycle emissions, but the answer is not as simple as tailpipe math. The real question is whether EVs still make sense when supply chains, critical minerals, war risk and energy systems all become less stable.

By Simran Bhimani

Systems analysis

Related paths

Shortcuts for deeper context after the article feed.

AI Power Demand Guide Data Center Grid Costs Guide Full archive