US Energy Grid Research — Session Transcript
📖 Platform Overview — What Every Section Does & When To Use It
This platform synthesizes US electricity grid data, interconnection policy, and energy economics into a single research environment for IHS scholars and policy analysts. The AI Research Agent (below) answers questions across all data. Navigate using the tabs in the header or the cards below.
⚡ Load Forecasting
Use when: you need a forward-looking demand/price/generation view for a specific ISO. Persistence and ensemble models project 24–168 hours ahead with confidence bands. Export results to CSV/JSON for external analysis.
🗺 Grid Infrastructure Maps
Use when: you need to see where infrastructure is — transmission lines, DER installations, data centers, or active regulatory filings — overlaid on ISO region boundaries. Six switchable layers; click any point for sourced data.
📊 Grid Utilization Analysis
Use when: you want to understand how "full" a region's grid is and what the economic case is for demand flexibility. Brattle's Untapped Grid framework models generation, transmission, and distribution utilization with consumer rate impact scenarios.
⬇ Data Downloads
Use when: you need raw data for external analysis. Downloads the in-memory ISO time series as CSV (all fields), extracts LMP price histories per operator, and exports grid topology (HIFLD transmission lines, planning zones) as CSV files.
🔌 Interconnection Queue
Use when: you need to understand the pipeline of projects waiting to connect — who's queued, how long the wait is, how many withdraw, and how costs have changed. LBNL Queued Up 2026 + ERCOT/PJM verified figures. Key for data center siting and policy analysis.
🔧 Flexibility Parameters
Use when: you need to understand what flexibility commitments regulators require of large loads — curtailment %, duration, frequency, ramp speed. Covers Brattle's 4-parameter framework, active ISO policies (TX SB 6, PJM CIFP, CPUC), and BYOG/BYOC structures.
💲 Cost Allocation & Rate Impact
Use when: you need to understand who pays for network upgrades and how that flows to consumer bills. Four allocation models (crediting, participant funding, Tx provider pays, co-location). Verified PJM capacity cost figures. Critical for any data center siting policy paper.
🏗 Network Upgrade Costs
Use when: you need $/kW cost data for interconnection by resource type and era. Historical LBNL baselines show why 77% of queue applicants withdraw. The co-location reform section tracks FERC's December 2025 rule and its impact on large-load economics.
⚖️ State Regulatory Activity
Use when: you need to track active dockets and legislation — BYOG/BYOC filings, IRP proceedings, rate cases, and queue reform compliance. All dockets are verified against primary sources. Filter by status (Enacted/Pending/Proposed) or filing type.
🎓 Energy Grid Learning Lab
Use when: you need orientation on energy grid concepts — generation mix, duck curves, interconnection, utilization, rate mechanics — at adjustable depth (College Student, Researcher, or Platform Guide). The Platform Guide mode provides tool-by-tool instructions for this platform.
🗺 State Electricity Rates
Use when: you need the current residential electricity rate (¢/kWh), average bill, and year-over-year trend for any state — mapped to ISO regions. Live EIA-861M data. Essential baseline for any ratepayer impact argument or legislative testimony.
🔬 Research Gap Intelligence
Use when: you need to identify where to focus IHS research. K-means clustering surfaces states where grid criticality outpaces research coverage — the "dark matter" of energy policy. Generates Gemini-powered story pitches framed through a classical liberal lens with sourced research questions.
AI Research Agent: The chat box below queries Gemini 2.0 Flash with full context of all loaded grid data. Ask it to compare ISOs, explain terms, model scenarios, or synthesize policy implications. Export the conversation as TXT, MD, CSV, JSON, or HTML Report using the toolbar below the chat. Per-message copy buttons appear on hover.
AI
Ask the Grid Research Agent
AI assistant trained on 10 US grid regions, energy reports, and policy data
K-Means
Z-Score Anomaly Detection
AI Story Pitches
50 States · 6 Dimensions
Data freshness: All facility counts, queue figures, and market data reflect the datasets loaded for this session. Counts may differ across sessions if underlying data files are updated. Always note the session date when citing specific numbers. Export any result with a timestamp using the chat export tools.
Session Transcript
0 exchanges
No Q&A yet this session
Research Gap Intelligence
K-means cluster analysis across all 50 states — surfaces under-researched, high-stakes energy policy areas
HOW TO USE THIS TOOL
Finding the stories and policy gaps that matter
① Run Gap Analysis
Click Run Gap Analysis. K-means clustering runs on all 50 states (+ DC) across 6 dimensions simultaneously. Takes 2–5 seconds in-browser — no data leaves your machine. Choose 5 clusters (default) for most research use. Use 6 if you want finer regional granularity; 4 for a high-level overview.
② Read the Scatter Plot
The chart plots every state by Grid Criticality (x-axis, higher = more stressed grid) vs Research Coverage (y-axis, higher = more studied). States in the bottom-right quadrant are the target: high stakes, low attention. Bubble size = ratepayer dollars at stake. State abbreviations are labeled directly on each bubble.
③ Prioritize by Gap Score
The Top Research Gaps panel ranks all states by Gap Score = Criticality − Coverage. Red scores (+35 and above) are the highest-priority gaps — states where grid stress is severe but research and policy attention is minimal. Use the Story Focus filter to narrow to a specific theme before generating pitches.
④ Generate Story Pitches
After running analysis, click Generate Story Pitches. Gemini 2.0 Flash synthesizes the cluster data into targeted research angles and policy narratives framed through a classical liberal lens — emphasizing market mechanisms, consumer protection, and regulatory capture. Each pitch includes a rationale, relevant states, and suggested sources.
Understanding the scores — what they measure and why they matter
Grid Criticality Index (0–100)
A composite of how much stress the state's grid is under. Built from: demand growth rate, weather vulnerability (heat/cold extremes), renewable integration complexity, ISO market structure, and data center load growth (pulled from dc_data.js when loaded). Above 70 = high stress. Texas (91), California (82), Arizona (81) lead. Higher criticality means grid failures, rate spikes, and policy fights are more likely — and more consequential.
Research Coverage Index (0–100)
A proxy for how much scholarly and regulatory attention the state receives. Derived from: active PUC docket volume, academic publication density, ISO regulatory sophistication, and ratepayer advocacy presence. California (91) and DC (78) are over-indexed — researched far more than their grid stress warrants. West Virginia (27), Arkansas (28), Alaska (22) are nearly invisible in the literature despite real grid challenges.
Gap Score = Criticality − Coverage
The core signal. A state with Criticality 74 and Coverage 36 has a gap of +38 — meaning it's under-researched relative to how much its grid matters. Gap > 35 (red) = research dark matter — high priority for scholars, journalists, and policymakers. Gap 18–35 (amber) = emerging gap, worth tracking. Negative gap = over-indexed, receiving more attention than grid stress warrants.
Ratepayer Exposure ($B/yr)
Total annual residential and commercial electricity spending in that state — a measure of how many people and how many dollars are affected if the grid fails or rates spike. Bubble size on the scatter plot. California ($41.8B) and Texas ($36.4B) dominate. A high-gap, high-exposure state like Virginia ($10.8B, gap ~+23) represents a large underserved research opportunity with major consumer stakes.
K-Means Clustering — What it does
Groups states into k clusters based on similarity across all 6 dimensions simultaneously (not just crit/cov). States in the same cluster share grid profiles — useful for comparative policy analysis. A researcher studying Arkansas can find structurally similar states in the same cluster and apply lessons across them. The algorithm runs locally in your browser in-browser — results may vary slightly between runs due to random initialization.
Quadrant Guide — Where to look
Bottom-right: Research Dark Matter — high criticality, low coverage. Richest source of original research.
Top-right: Well-Studied Critical — high stakes, appropriate attention. Still worth covering but crowded.
Top-left: Over-Indexed — modest grid stress, outsized attention. Policy narratives here may be distorted.
Bottom-left: Low-Stakes Background — stable grids, low coverage is appropriate.
Top-right: Well-Studied Critical — high stakes, appropriate attention. Still worth covering but crowded.
Top-left: Over-Indexed — modest grid stress, outsized attention. Policy narratives here may be distorted.
Bottom-left: Low-Stakes Background — stable grids, low coverage is appropriate.
Clusters (k)
Story focus
Ready — click Run Gap Analysis to begin k-means clustering
HOW TO USE
1
Set clusters & focus — Choose 5 clusters (default) for most use. Use "Story focus" to narrow to dark matter gaps, grid transition states, or ratepayer exposure before generating pitches.
2
Click Run Gap Analysis — K-means clusters all 50 states across 6 grid dimensions in ~3 seconds, entirely in-browser. No data leaves your machine.
3
Read the scatter plot — Each bubble is a state. Bottom-right = high-stakes, under-studied (your target). Bubble size = ratepayer dollars at risk. State codes label each bubble.
4
Generate Story Pitches — Gemini 2.0 Flash synthesizes the cluster findings into targeted research narratives with classical liberal framing, source suggestions, and policy angles.
READING THE SCORES
CRITICALITY
How stressed the grid is — demand growth, weather risk, renewable complexity, data center load. 70+ = high stress.
COVERAGE
How much research & regulatory attention the state gets — PUC dockets, publications, advocacy. Low = under-studied.
GAP SCORE
Criticality minus Coverage. +35 or more = dark matter — high priority. 18–35 = emerging gap. Negative = over-indexed.
BUBBLE SIZE
Annual ratepayer electricity spend ($B/yr) — how many consumer dollars are at stake if this grid goes wrong.
Quadrant guide
↘ Bottom-right
Research Dark Matter — high stakes, ignored. Start here.
↗ Top-right
Well-Studied Critical — important but crowded field.
↖ Top-left
Over-Indexed — more coverage than the grid risk warrants.
↙ Bottom-left
Low-Stakes Background — stable grids, low coverage is fine.
Criticality vs. Coverage Matrix
Bottom-right = high-stakes, under-studied · bubble size = ratepayer exposure
Top Research Gaps
0 gaps
Run analysis to see ranked gaps
Cluster Profiles
Multi-dimensional state groupings sorted by average gap score
Run analysis to see cluster profiles
AI Research Story Pitches
What scholars and journalists are missing — with classical liberal framing
Run gap analysis, then click Generate Story Pitches
Live Anomaly Detection
0 anomalies
Click Scan ISO Data to run z-score detection on loaded grid time series (threshold: z > 2.5)
Load Forecasting
Persistence-based forecast model with conformal prediction intervals
📖 How to Use Load Forecasting
Select an ISO/RTO, choose what to forecast (demand, generation, price, or utilization), set your horizon and model type, then click Run Forecast. The chart shows recent actuals alongside the forward projection and a shaded confidence band.
Model Types
Same-Hour-Yesterday (Persistence) — uses yesterday's same-hour value as tomorrow's forecast. Simple and often surprisingly accurate for stable grids.
Same-Hour-Last-Week — uses the same hour 7 days ago, better for capturing weekly cycles (weekday vs. weekend).
Ensemble Average — averages both models, reducing individual model error.
Same-Hour-Last-Week — uses the same hour 7 days ago, better for capturing weekly cycles (weekday vs. weekend).
Ensemble Average — averages both models, reducing individual model error.
Confidence Band
The shaded area around the forecast line is a conformal prediction interval — it widens as the forecast horizon grows because uncertainty compounds over time. At 90% coverage, 9 of 10 actual values should fall inside the band. Higher confidence (95%) = wider band; lower (80%) = tighter but less reliable.
Performance Metrics
RMSE (Root Mean Squared Error) — average forecast miss in MW/MWh. Lower = better. Penalizes large errors more than small ones.
MAPE (Mean Absolute Percentage Error) — average miss as a % of actual. <3% is excellent, <5% is good for day-ahead.
Coverage — % of actual values that fell inside the confidence band. Should be close to your chosen CI (e.g., 90%).
Skill Score — how much better this model is vs. a naïve "flat line" baseline. Positive = better than guessing.
MAPE (Mean Absolute Percentage Error) — average miss as a % of actual. <3% is excellent, <5% is good for day-ahead.
Coverage — % of actual values that fell inside the confidence band. Should be close to your chosen CI (e.g., 90%).
Skill Score — how much better this model is vs. a naïve "flat line" baseline. Positive = better than guessing.
Export Options
After running a forecast, an Export toolbar appears below the chart. CSV downloads the full forecast table (timestamp, actual, forecast, lower bound, upper bound). JSON includes model metadata. Copy pastes the table to clipboard. PDF/Print generates a printable report.
Data source: EIA-930 Hourly Grid Monitor (eia.gov/electricity/gridmonitor) via operator
*_data.js files. The model trains on the most recent 30 days of confirmed actuals. Future rows from ISO day-ahead schedules are stripped before training.
Metric to Forecast
⚡ Demand
🔋 Generation
💲 Price
📊 Utilization
Operator
Forecast Horizon
Model Type
Confidence Interval
Model Performance
RMSE
—
MAPE
—
Coverage
—
Skill Score
—
Demand Forecast vs Actual — ERCOT
Prior 24h Actual
Forecast
Confidence Band
Scroll to zoom · Drag to pan
Grid Infrastructure Maps
Real US transmission topology from HIFLD Open Data (DHS/CISA) — 3,400+ HV lines ≥345 kV. Planning zones from ISO/RTO operational boundaries. Research reference: TAMU ACTIVSg Series25 (2025)
📖 Map Layers & Controls
🗺 ISO Regions
States shaded by which ISO/RTO operates their grid. Useful for understanding market boundaries — generation and pricing rules differ completely between regions.
⚡ Transmission Network
Real high-voltage lines from HIFLD (DHS/CISA) — only lines ≥345 kV are shown. Color = voltage level: yellow=345 kV, orange=500 kV, red=765 kV. Higher voltage = more power moved over longer distances.
☀ DER Research Map
Distributed Energy Resources fetched live from EIA Form 860 (api.eia.gov). Shows solar, storage, VPPs, wind, fuel cells, and microgrids. Filter by type. Click a dot for capacity (kW), annual MWh, interconnection queue status, and CO₂ avoided.
🏢 Data Centers & GETs
7,060+ US facilities from OSM/HIFLD/EIA-860. GETs = Grid-Enhancing Technologies (advanced conductors, topology optimization, flow control) — FERC-tracked equipment that can unlock 15–17% additional transmission capacity on existing lines without building new infrastructure.
⚖️ Regulatory Filings
Active FERC and state PUC dockets mapped to location. Types: BYOG/BYOC (data centers building own generation for fast-track interconnection), Queue Reform (FERC Order 2023 compliance filings), Utility IRP (Integrated Resource Plan — long-term capacity planning), Rate Case (utility seeking a rate change from the PUC).
🔍 Navigation & Filters
Use the ISO Filter chips to isolate a region. +/− buttons zoom; ⌖ resets. Scroll to zoom, drag to pan. The stats bar (top of map) updates with counts for the active layer. Hover any element for a tooltip with source citation.
Data sources: Transmission lines — HIFLD Open Data (hifld-geoplatform.opendata.arcgis.com) · DER — EIA Form 860 (api.eia.gov) · Planning zones — ISO/RTO operational boundaries · Facilities — dc_data.js (OSM + HIFLD + EIA-860 + EPA)
96
HV Lines Shown
76
Planning Zones
38
Utility Zones
82K
Total Buses (Model)
Data Source: Texas A&M University
Real transmission topology: HIFLD Open Data (hifld-geoplatform.opendata.arcgis.com)
Planning zones: ISO/RTO operational boundaries · EIA Form 860 (2026)
Research ref: TAMU ACTIVSg Series25 (electricgrids.engr.tamu.edu)
Visualization © IHS Energy Grid Research · EIA-930
Real transmission topology: HIFLD Open Data (hifld-geoplatform.opendata.arcgis.com)
Planning zones: ISO/RTO operational boundaries · EIA Form 860 (2026)
Research ref: TAMU ACTIVSg Series25 (electricgrids.engr.tamu.edu)
Visualization © IHS Energy Grid Research · EIA-930
Grid Utilization Analysis
Based on Brattle Group "The Untapped Grid" (March 2026) — GridLab and Utilize Coalition. Utilization baselines derived from FERC Form 714 (2022 to 2024 average). Rate impact model is illustrative; actual outcomes depend on local conditions.
📖 How to Read Grid Utilization Analysis
The US grid carries far less power than it physically could. Brattle's March 2026 analysis found that demand flexibility and smarter scheduling — without new wires or generators — could absorb 175 GW of new load by 2030 while lowering consumer rates by 3.4% versus a status-quo baseline. This section models that opportunity across three infrastructure layers.
⚡ Generation Utilization
Formula: Output (MWh) ÷ (Unforced Capacity × 8,760 hrs). US average: ~50%. This isn't inefficiency — generators are intentionally held in reserve for peak demand and system reliability. The question is whether flexibility tools can reduce how much reserve margin is needed, freeing more of that latent capacity for real loads. ELCC-weighted means variable renewables are credited at their Effective Load-Carrying Capability, not nameplate.
🔌 Transmission Utilization
Formula: Actual flows ÷ Firm Transfer Capability. Naturally lower than generation — N-1 security requires that the grid remain stable even if any single line fails, so lines can't run at 100%. GETs (Grid-Enhancing Technologies) — advanced power flow control, dynamic line ratings, topology optimization — can unlock 15–17% additional transfer capacity on existing wires without construction. This is Brattle's primary transmission opportunity.
🏘 Distribution Utilization
Formula: Feeder load ÷ Thermal loading limits. The most constrained layer for electrification — EV charging and heat pumps cluster geographically, creating local peak load events that overwhelm neighborhood-level circuits. ~70% of US transmission lines are over 25 years old. Smart panels, flexible interconnection standards, and managed charging can shift peak loads to unlock headroom without replacing the wire.
📊 Scenario Controls
ISO/Operator — selects the region. Baselines differ significantly: BPA (hydro-dominated) vs. ERCOT (wind-heavy) vs. ISO-NE (constrained, high-cost).
Load Growth Scenario — 20/25/30% growth by 2030, driven primarily by data centers. Brattle's base case is 25%.
DER Cost ($/kW-yr) — annualized cost of deploying distributed energy resources. $50 = Brattle base; $25 = optimistic (favorable VPP contracts); $100 = conservative (legacy utility programs).
Strategy — Status Quo adds new infrastructure; Utilization Focus maximizes existing assets; Extreme avoids all new builds.
Load Growth Scenario — 20/25/30% growth by 2030, driven primarily by data centers. Brattle's base case is 25%.
DER Cost ($/kW-yr) — annualized cost of deploying distributed energy resources. $50 = Brattle base; $25 = optimistic (favorable VPP contracts); $100 = conservative (legacy utility programs).
Strategy — Status Quo adds new infrastructure; Utilization Focus maximizes existing assets; Extreme avoids all new builds.
📈 KPI Cards (Top Row)
The six stat cards update when you change any scenario parameter:
Generation Util. — % of unforced capacity actually used.
Tx Utilization — % of firm transfer capability used.
Dist. Utilization — % of feeder loading limit used.
Rate Impact — change vs. a no-load-growth baseline.
10-Yr Bill Savings — national consumer savings from the utilization strategy vs. status quo.
Connection Time — estimated time to connect new load under this strategy.
Generation Util. — % of unforced capacity actually used.
Tx Utilization — % of firm transfer capability used.
Dist. Utilization — % of feeder loading limit used.
Rate Impact — change vs. a no-load-growth baseline.
10-Yr Bill Savings — national consumer savings from the utilization strategy vs. status quo.
Connection Time — estimated time to connect new load under this strategy.
🔧 Utilization Tools (DER Catalog)
The right-side panel lists Brattle's catalog of tools that improve utilization. VPP (Virtual Power Plant) — aggregates distributed resources (rooftop solar, batteries, smart thermostats) into a dispatchable block. BYODC — data centers investing in residential DER in exchange for grid headroom at their site. GETs — software and hardware upgrades on existing transmission lines. Flex Interconnection — connects new loads at lower cost by accepting occasional curtailment.
Affordability Index (bottom table): Ranks all ISO/RTOs by a composite of utilization, load growth, rate trend, and DER opportunity. High affordability pressure = region where rate increases are most likely without intervention — highest priority for research and policy attention. Cross-reference with State Electricity Rates for current baselines and Research Gap Intelligence for scholarly coverage gaps.
Source: Brattle Group "The Untapped Grid" (Mar 2026), commissioned by GridLab & Utilize Coalition · FERC Form 714 (2022–2024 avg) · LBNL capacity market data
ISO / Operator
Load Growth Scenario
DER Cost ($/kW-yr)
Strategy
Generation Util.
—
% avg capacity
Tx Utilization
—
% firm transfer cap
Dist. Utilization
—
% feeder loading
Rate Impact
—
vs. no-growth base
10-Yr Bill Savings
—
national estimate
Connection Time
—
to connect new load
Capacity Utilization by ISO — Generation Layer
Brattle FERC Form 714 · 2022–2024 avg
Rate Impact: Status Quo vs. Utilization Focus
Brattle Scenario Model
■ Status Quo (+1.4% rate)
■ Utilization Focus (-3.4% rate)
■ No Load Growth (baseline)
National Consumer Bill Savings Potential — 10-Year Horizon
Utilization Formula
Utilization (%) =
Total energy delivered (MWh)
─────────────────────────
Available capacity (MW) × 8,760 hrs
⚡ Generation
Output ÷ unforced capacity (ELCC-weighted). US avg: ~50%. Peak reserve margin reduces ceiling.
🔌 Transmission
Flows ÷ firm transfer capability. Naturally lower than gen — N-1 redundancy required. GETs can unlock 15–17% additional capacity.
🏘 Distribution
Feeder load ÷ loading limits. Most constrained layer for EV / heat pump rollout. Smart panels + flex interconnection target this tier.
Utilization Tools (Brattle Catalog)
Key Benchmarks
National DER flexibility potential200 GW
US peak load growth by 2030+175 GW
US peak load growth by 2035+270 GW
10-yr rate swing (util. focus)−4.8%
VPP cost vs. conventional plant40–60% less
Tx lines over 25 years old~70%
DER capacity credit — generation90%
DER capacity credit — transmission25%
DER capacity credit — distribution40%
Base DER net cost (util. base case)$50/kW-yr
Source: Brattle Group, "The Untapped Grid," March 2026. Commissioned by GridLab & Utilize Coalition.
Affordability Pressure Index — All ISO/RTOs
Utilization · Load Growth · Rate Trend
| ISO/RTO | Gen Util. (est.) | Tx Util. (est.) | Peak GW | 2030 Load Growth | Headroom Est. | Rate Trend | Affordability Pressure | Top DER Opportunity |
|---|
Data Downloads
📖 What Can Be Downloaded & How
Complete Operator Datasets (top section)
All fields from the operator's hourly time series — demand (MW), net generation by fuel type, interchange, and where available: LMP price, reserve margin, and frequency. Downloaded directly from whichever data is currently loaded in memory. To populate: go to the Load Forecasting tab, select the operator, click Run Forecast, then return here to download. The button will build the CSV from live data.
LMP Price Data (bottom section)
LMP = Locational Marginal Price ($/MWh) — the real-time wholesale price of electricity at a specific node. It reflects generation cost + transmission congestion + marginal losses. Note: TVA, BPA, and FPL use administered rates, not LMP-based market clearing — their "price" column reflects posted tariff rates, not nodal auction prices. All Operators snapshot gives one row per operator with the latest price, record count, and date range.
Grid Topology Data (HIFLD)
Downloads three CSVs simultaneously: hifld_transmission_lines.csv (3,400+ real HV lines ≥345 kV with voltage, endpoints, and coordinates), tamu_planning_zones.csv (76 ISO planning zones with geographic boundaries), and utility_zones.csv (38 utility service territory definitions). Sourced from HIFLD Open Data (DHS/CISA) and TAMU ACTIVSg Series25.
Chat Session Exports (Dashboard tab)
The Q&A agent session can be exported from the Dashboard tab in TXT, MD, CSV, JSON, or HTML Report formats. Each export includes a timestamp, message count, and a data-provenance note. Use for citations, research documentation, or internal reporting.
All downloads generate files client-side — no data is sent to any server. Timestamps in filenames reflect the moment of download.
ERCOT Complete Dataset
Electric Reliability Council of Texas — 5 year hourly data
PJM Complete Dataset
PJM Interconnection — 5 year hourly data
MISO Complete Dataset
Midcontinent ISO — 5 year hourly data
SPP Complete Dataset
Southwest Power Pool — 5 year hourly data
CAISO Complete Dataset
California ISO — 5 year hourly data
NYISO Complete Dataset
New York ISO — 5 year hourly data
ISO-NE Complete Dataset
ISO New England — 5 year hourly data
TVA Complete Dataset
Tennessee Valley Authority — 5 year hourly data
BPA Complete Dataset
Bonneville Power Administration — 5 year hourly data
FPL Complete Dataset
Florida Power & Light — 5 year hourly data
Grid Topology Data (HIFLD)
Real US transmission lines from HIFLD Open Data + ISO/RTO zones
LMP Price Data
Locational Marginal Price ($/MWh) from loaded operator data files. Cards show a warning when the corresponding
*_data.js file is not present.All Operators — Price Snapshot
Current $/MWh for all 10 operators (7 ISOs + TVA + BPA + FPL) in one CSV — note: TVA/BPA/FPL use administered rates, not LMP
ERCOT — LMP History
Hourly LMP price series from loaded data file
PJM — LMP History
Hourly LMP price series from loaded data file
CAISO — LMP History
Hourly LMP price series from loaded data file
NYISO — LMP History
Hourly LMP price series from loaded data file
MISO — LMP History
Hourly LMP price series from loaded data file
SPP — LMP History
Hourly LMP price series from loaded data file
ISO-NE — LMP History
Hourly LMP price series from loaded data file
TVA — LMP History
Hourly LMP price series from loaded data file
BPA — LMP History
Hourly LMP price series from loaded data file
FPL — LMP History
Hourly LMP price series from loaded data file
Large Load Interconnection Queue
Verified data from ERCOT (Q1 2026), LBNL Queued Up 2026 edition (data through Q1 2026), and PJM 2026 Long Term Load Forecast. Supplemented by EIA-860 Form data (2026) and FERC Order 2023 cluster reform tracking. Figures are point in time and may not reflect subsequent queue additions or withdrawals.
📖 How to Read This Section
This section tracks what happens before a data center (or any large load) can draw power from the grid. Every new facility must enter the interconnection queue — a multi-year study process managed by the regional grid operator (ISO/RTO). The queue determines whether the grid can physically and safely support the new load, and what upgrades are needed.
Key Terms
Interconnection QueueThe backlog of projects (generation or load) waiting for grid connection studies. Think of it as a "waiting list" to plug into the grid.
Large Load QueueThe subset of queue entries that are demand-side (data centers, manufacturing, etc.) rather than generation (solar, wind, gas plants).
Energized (MW)Capacity that has completed all studies and is now live — actually drawing power from the grid. This is the "done" number.
DC Share (%)What fraction of queued or forecast load growth comes from data centers specifically. PJM's 94% means nearly all new demand is data centers.
GW QueuedGigawatts of requested capacity still waiting for interconnection approval. Higher = longer wait times for everyone.
Withdrawal RatePercentage of projects that entered the queue but gave up before completion — typically because costs or wait times became prohibitive.
Completion RatePercentage of projects that actually made it from queue entry to commercial operation. Only 13% historically.
Executed IA (GW)Projects that have signed an Interconnection Agreement — a binding contract that commits both parties. Much closer to "real" than a queue request.
IR → CODInterconnection Request to Commercial Operation Date — the full timeline from first filing to actually being energized.
$/kW (IC Cost)Interconnection cost per kilowatt — what the developer pays for grid upgrades needed to support their project. Higher = more expensive to connect.
CIFP / BYOGCritical Issue Fast Path (PJM expedited review) and Bring Your Own Generation (loads that co-locate with on-site generation for faster approval).
FERC Order 2023Federal rule reforming the interconnection process: cluster-based studies, financial readiness deposits, faster timelines. First batches processing mid-2026.
What the Cards & Charts Show
⬛ Stat Cards (top)National-level queue snapshot from LBNL: total queue size, withdrawal rate, completion rate, median wait time, active projects, and projects with signed agreements.
⬛ ERCOT / PJM CardsISO-specific verified data. Amber left border = source-cited figures with provenance. Shows queue size, DC share, energized MW, wait times, and legislative/regulatory context.
🟩 PJM Cost ChartThree eras of PJM interconnection costs: green = pre-reform baseline ($29/kW, cheap), red = post-2020 reform era ($240/kW, 8× increase), purple = withdrawn projects ($700/kW — projects that quit because costs were too high).
📊 Growth ChartUS total generation queue by year (LBNL annual reports). Amber bars = total GW in queue. Green bars = subset with executed Interconnection Agreements (signed commitments). The gap between them shows speculative vs. committed capacity.
How to Use This
For policy research: Compare queue backlogs to energized MW to see the "conversion funnel" — how much of queued demand actually gets built. The 77% withdrawal rate means most projects never happen.
For siting analysis: ERCOT's 233 GW queue vs. 5.3 GW energized reveals a ~44:1 ratio of demand to delivered capacity. PJM's 8.5+ year wait time sets the planning horizon.
For cost modeling: The PJM cost chart shows interconnection is no longer cheap infrastructure — at $240–700/kW, it rivals generation costs and drives project economics.
Cross-reference: Use the Grid Utilization tab to see where spare capacity exists (potentially shorter queues), and Network Upgrade Costs for detailed $/kW breakdowns by resource type.
For siting analysis: ERCOT's 233 GW queue vs. 5.3 GW energized reveals a ~44:1 ratio of demand to delivered capacity. PJM's 8.5+ year wait time sets the planning horizon.
For cost modeling: The PJM cost chart shows interconnection is no longer cheap infrastructure — at $240–700/kW, it rivals generation costs and drives project economics.
Cross-reference: Use the Grid Utilization tab to see where spare capacity exists (potentially shorter queues), and Network Upgrade Costs for detailed $/kW breakdowns by resource type.
PJM Network Upgrade Cost — Reform Era ImpactLBNL 2023+2026
📄 LBNL PJM IC Cost Analysis (Jan 2023): $29/kW pre-2020 → $240/kW 2020-2022. Non-ISO BAs: $194/kW complete projects 2018-2024 (LBNL Feb 2026). Withdrawn projects: $599/kW.
US Total Generation Queue — All Resource Types (LBNL Annual Reports)
📄 LBNL Queued Up annual reports (emp.lbl.gov/queues). Q1 2026 active queue ~1,420 GW — dominated by solar+storage and data center large loads. ERCOT's large-load queue: 233 GW (Dec 2025), growing through Q1 2026. FERC Order 2023 cluster reforms processing first batches; early results expected mid-2026.
Flexibility Commitment Parameters
Brattle "Untapped Grid" (Mar 2026, updated) 4-parameter framework — and active ISO policies requiring flexibility as a condition of interconnection
📖 What Is Flexibility & Why Does It Matter?
Grid operators now require large new loads (especially data centers) to commit to operating flexibly — reducing consumption during grid stress events — in exchange for faster interconnection and lower costs. This shifts the conversation from "build more supply" to "manage demand smarter." The Brattle Group's March 2026 report defined a 4-parameter framework now being adopted by regulators.
Curtailment % — How Much to Cut
The fraction of load a facility must shed during a grid stress event. Example: 25% curtailment on a 200 MW data center = drop to 150 MW. Brattle's baseline: 10–25%. ERCOT SB 6 mandates 75 MW+ facilities register with curtailment capability.
Duration — How Long Each Event
How many consecutive hours curtailment must be sustained per event. Brattle baseline: 2–4 hours. This is short enough that AI training workloads can pause without data loss, and batched compute can be rescheduled to off-peak windows.
Frequency — Total Hours Per Year
Total annual hours the facility may be called upon to curtail. Brattle baseline: 100–200 hours/year (~1.1–2.3% of the year). This bounds the business impact and gives load owners confidence curtailment won't be invoked frivolously.
Ramp Time — How Fast to Respond
Time from curtailment signal to full compliance. Brattle baseline: 15–60 seconds. Modern data center power management systems can shed load in <1 second. Faster ramp = more valuable for frequency regulation, potentially earning capacity market payments.
BYOG / BYOC — Build Your Own
BYOG = Bring Your Own Generation. A large load co-locates with its own generator (e.g., gas turbine, solar+storage). PJM's CIFP track: if a 1,000 MW generator serves a 900 MW data center, only the net 100 MW injection is studied — dramatically shortening queue time. BYOC = Bring Your Own Capacity (California variant under CPUC R.24-01-018).
Demand Response (DR)
Programs that pay consumers to reduce usage during peaks. Different from mandatory flexibility — DR is voluntary and compensated. ERCOT has ~12,550 MW of DR capacity; PJM ~18,500 MW. Data centers participating in DR programs can partially offset interconnection costs through capacity market payments.
Sources: Brattle Group "The Untapped Grid" (Mar 2026) · NVIDIA DSX Flex framework (CERAWeek 2026) · ERCOT SB 6 (Jun 2025) · PJM CIFP (Dec 2025) · CPUC R.24-01-018 · FERC Co-Location Order (Dec 2025)
Brattle 4-Parameter FrameworkMar 2026
Active ISO / Regulatory Frameworks
Cost Allocation & Rate Impact
Verified PJM capacity cost figures and consumer impact estimates — fabricated per-ISO rate tables removed
📖 Understanding Cost Allocation & Rate Impact
When a large load connects to the grid, the question of who pays for the required upgrades is the most politically contentious issue in data center siting. The answer depends entirely on which cost allocation model applies in that ISO — and the difference between models runs into the billions of dollars for ratepayers.
4 Cost Allocation Models
Crediting (CAISO): Developer pays upfront, receives reimbursement as others use the upgraded line. Risk stays with developer early; recovers over time.
Participant Funding (PJM, MISO, SPP, NYISO, ISO-NE): Developer pays all upgrade costs, no reimbursement. Full cost internalized — drives project economics.
Tx Provider Pays (ERCOT): Upgrade costs spread across all ratepayers; developer bears only curtailment risk. Lowers developer cost but socializes expense.
Co-Location (PJM Dec 2025): Only the net grid injection from a co-located gen+load pair is studied. Can reduce interconnection cost by 80–90%.
Participant Funding (PJM, MISO, SPP, NYISO, ISO-NE): Developer pays all upgrade costs, no reimbursement. Full cost internalized — drives project economics.
Tx Provider Pays (ERCOT): Upgrade costs spread across all ratepayers; developer bears only curtailment risk. Lowers developer cost but socializes expense.
Co-Location (PJM Dec 2025): Only the net grid injection from a co-located gen+load pair is studied. Can reduce interconnection cost by 80–90%.
Capacity Cost & $/MW-day
Capacity market — pays generators to be available at peak demand, not for actual energy. PJM's capacity prices: $28.92/MW-day (2023–24) → $329.17/MW-day (2025–26) — an 11× increase driven primarily by data center load growth.
$/MW-day means dollars per megawatt of capacity per day. $329/MW-day × 365 = ~$120,000/MW-year. A 1,000 MW data center paying capacity costs faces ~$120M/year before any energy charges.
$/MW-day means dollars per megawatt of capacity per day. $329/MW-day × 365 = ~$120,000/MW-year. A 1,000 MW data center paying capacity costs faces ~$120M/year before any energy charges.
Wholesale vs. Retail Rate Spread
LMP (wholesale) — what generators earn per MWh at the bus. Set by hourly market clearing.
Retail rate — what households pay. Includes transmission, distribution, capacity, policy charges, utility margin, and taxes. ISO-NE's retail-to-wholesale spread is ~3× SPP's — ISO-NE ratepayers pay far more above the raw generation cost than SPP ratepayers.
Rate impact — how much a new interconnection changes an existing customer's bill. Depends on cost allocation model and whether upgrade costs are embedded in transmission tariffs.
Retail rate — what households pay. Includes transmission, distribution, capacity, policy charges, utility margin, and taxes. ISO-NE's retail-to-wholesale spread is ~3× SPP's — ISO-NE ratepayers pay far more above the raw generation cost than SPP ratepayers.
Rate impact — how much a new interconnection changes an existing customer's bill. Depends on cost allocation model and whether upgrade costs are embedded in transmission tariffs.
How to Read This Tab
The context card shows verified PJM capacity auction results with sourced figures. The Brattle note frames the consumer impact in Brattle's Untapped Grid model. The policy table shows each ISO's current cost allocation rule — enacted law vs. pending rulemaking vs. proposed reform. Use this to understand the regulatory landscape for a specific region before modeling rate impact.
Sources: PJM Capacity Auctions 2023–2026 · PJM IMM Nov 2025 · Synapse Energy Economics · Brattle Group (Mar 2026) · DOE/NREL i2X · FERC Co-Location Order (Dec 2025)
Cost Allocation Policy by ISO / Jurisdiction
What Are Ratepayers Paying?
EIA-861M residential electricity rates for all 50 states and D.C., mapped to 16 ISO/RTO and utility wholesale regions
📖 Reading the Ratepayer Map
¢/kWh — Retail Electricity Rate
What a typical household pays per kilowatt-hour — the all-in price including generation, transmission, distribution, taxes, and policy charges. National range: ~10¢ (WA, hydro-dominated) to ~38¢ (HI, isolated island grid). National average: ~18¢. This is what voters and legislators care about — not wholesale LMP.
Color Modes
Color by ISO Region — states colored by which grid operator manages their market. Useful for comparing policy environments.
Color by Rate — gradient from teal (cheap) to red (expensive). Quickly shows geographic rate disparities. The gradient legend below the map shows the actual min/max ¢/kWh values.
Color by Rate — gradient from teal (cheap) to red (expensive). Quickly shows geographic rate disparities. The gradient legend below the map shows the actual min/max ¢/kWh values.
State Detail Panel
Click any state to see: current rate (¢/kWh), average monthly bill ($), year-over-year change (same month prior year — controls for seasonality), and a 36-month sparkline showing the rate trend. A rising sparkline heading into data center construction periods is a political risk signal for siting decisions.
ISO Region Rankings
The right panel ranks ISO regions by average residential rate. High-rate regions (ISO-NE, NYISO) reflect congestion, fuel costs, and dense infrastructure. Low-rate regions (SPP, MISO) reflect cheap coal/wind generation and socialized transmission costs. Cross-reference with Grid Utilization to find regions with both spare capacity and low rates.
Source: EIA Form 861M — Monthly Electric Utility Report (api.eia.gov). Live fetch with cached fallback from ratepayer_data.js. Average bill = (utility revenue × 1,000) ÷ customer count.
Select a State
Click any state to view its residential rate, average bill, year-over-year change, and 36-month trend.
ISO Region Rankings¢/kWh
Network Upgrade Cost Tracking
Verified $/kW interconnection costs — LBNL 2023 historical baselines + LBNL Feb 2026 non-ISO BA analysis + FERC Dec 2025 co-location reforms; four cost allocation models
📖 Understanding Network Upgrade Costs
Before any large load or generator can connect to the grid, the ISO conducts interconnection studies to determine what infrastructure must be upgraded to accommodate the new connection. These network upgrade costs have exploded since 2020 — they are now the dominant driver of interconnection economics and the primary reason 77% of projects withdraw from the queue.
$/kW — The Core Metric
Interconnection cost expressed per kilowatt of project capacity. Pre-2020 PJM baseline: $29/kW — considered cheap infrastructure. Post-2020 active projects: $240/kW — an 8× increase. Withdrawn projects: $599/kW — projects that quit because upgrade costs were uneconomic. Non-ISO BAs (utilities outside major ISOs): $194/kW for completed projects (LBNL Feb 2026).
POI Cost vs. Network Upgrade Cost
POI (Point of Interconnection) cost — equipment right at the connection point: breakers, transformers, metering. Historically stable, $10–30/kW.
Network upgrade cost — improvements to the broader transmission system miles away from the site. Can trigger $100M+ in upgrades. This is what has exploded. A solar farm may need a new substation; a 500 MW data center may require miles of new 345 kV line.
Network upgrade cost — improvements to the broader transmission system miles away from the site. Can trigger $100M+ in upgrades. This is what has exploded. A solar farm may need a new substation; a 500 MW data center may require miles of new 345 kV line.
Resource Type Cost Differences
Interconnection costs vary significantly by resource type. Large Loads (data centers) typically face the highest costs because they need firm power — no curtailment. Storage can be sited more flexibly. Solar often has lower network upgrade costs but higher withdrawal rates due to queue congestion. The resource type table in this section shows LBNL's $/kW breakdown by category.
FERC Co-Location Reform (Dec 2025)
FERC's December 2025 co-location order allows a generator and co-located data center to be studied as a net injection — only the surplus power exported to the grid triggers network upgrade studies. A 1,000 MW gas plant serving a 950 MW data center = only 50 MW of net injection studied. PJM compliance filing submitted Feb 2026. DOE proposed making this nationwide by April 2026.
Sources: LBNL PJM IC Cost Analysis (Jan 2023) · LBNL Non-ISO BA Analysis (Feb 2026) · DOE/NREL i2X Initiative · FERC Co-Location Order (Dec 2025) · Brattle Group (Mar 2026)
PJM Interconnection Cost by Resource TypeLBNL 2023+2026
Cost Allocation ModelsDOE/NREL i2X + FERC 2025
State-Level Regulatory Activity
Verified proceedings only — dockets confirmed via Brattle, Synapse, White & Case, and primary sources; fabricated docket numbers removed
📖 Reading the Regulatory Activity Table
Status Badges
Enacted — signed into law or final rule issued. Binding.
Pending — open docket or bill awaiting vote/order. Active regulatory risk.
Proposed — filed but not yet in active proceeding. Early-stage signal.
Use the filter buttons to focus on one status type.
Pending — open docket or bill awaiting vote/order. Active regulatory risk.
Proposed — filed but not yet in active proceeding. Early-stage signal.
Use the filter buttons to focus on one status type.
Filing Types
BYOG/BYOC — Bring Your Own Generation/Capacity filings. Data centers seeking co-location or on-site generation approval for expedited interconnection.
Queue/Tariff Reform — FERC Order 2023 compliance filings and ISO tariff updates restructuring the interconnection queue.
Utility IRP — Integrated Resource Plan. A utility's long-term (10–20 year) capacity plan filed with the PUC; increasingly dominated by data center load growth projections.
Rate Case — utility petition to the PUC to change rates or tariffs. Often triggered by large load additions or infrastructure investments.
Queue/Tariff Reform — FERC Order 2023 compliance filings and ISO tariff updates restructuring the interconnection queue.
Utility IRP — Integrated Resource Plan. A utility's long-term (10–20 year) capacity plan filed with the PUC; increasingly dominated by data center load growth projections.
Rate Case — utility petition to the PUC to change rates or tariffs. Often triggered by large load additions or infrastructure investments.
Table Columns
Docket/Bill — the official case number or bill number. Use this to search FERC eLibrary, state PUC dockets, or state legislature websites for original filings.
Key Provision — the most consequential clause for grid access or consumer cost.
Updated — date of last confirmed status change. Verify before citing; regulatory proceedings move quickly.
Source — primary citation. Only verified dockets are included.
Key Provision — the most consequential clause for grid access or consumer cost.
Updated — date of last confirmed status change. Verify before citing; regulatory proceedings move quickly.
Source — primary citation. Only verified dockets are included.
Proceedings Chart
The donut chart on the right shows the mix of proceeding types currently active. A high share of IRP filings signals utilities are actively revising capacity plans due to load growth. A high share of Rate Cases signals costs are being passed to consumers. Use this to quickly read the regulatory temperature of a region before drafting a research pitch.
Sources: TX SB 6 · CPUC R.24-01-018 · PA PUC M-2025-3054271 · VA SB 1498 · NJ AB 5564 · UT SB 132 · FERC Order 2023 · PJM Board Jan 2026 · Brattle (Mar 2026) · Synapse Energy Economics · White & Case
Active Proceedings & Legislation
| State | ISO | Docket/Bill | Topic | Status | Key Provision | Updated | Source |
|---|
Proceedings by Type
📄 Sources: TX SB 6; CPUC R.24-01-018; PA PUC M-2025-3054271; VA SB 1498; NJ AB 5564; UT SB 132; FERC Order 2023; PJM Board Jan 2026
Energy Grid Learning Lab
Interactive educational modules on how the US power grid works — generation mix, duck curves, interconnection, utilization, and the policy levers that shape what ratepayers pay. Use Platform Guide mode for tool-by-tool research guidance.
📖 How to Use the Learning Lab
🎓 Audience Modes
College Student — conceptual explanations with everyday analogies. Suitable for a first encounter with grid concepts, generation mix, or interconnection.
Researcher — adds quantitative depth: capacity factors, LMP mechanics, queue cost drivers, utilization formulas. For scholars and policy analysts.
Platform Guide — step-by-step instructions for using every tool in this application. Start here if you're new to the platform.
Researcher — adds quantitative depth: capacity factors, LMP mechanics, queue cost drivers, utilization formulas. For scholars and policy analysts.
Platform Guide — step-by-step instructions for using every tool in this application. Start here if you're new to the platform.
⚡ Generation Mix Bar
Live US generation mix fetched from EIA-930. Each colored segment is a fuel type — width = share of total output right now. Baseload (nuclear, hydro) runs flat 24/7. Dispatchable (gas peakers) ramp up and down with demand. Variable renewables (wind, solar) fluctuate with weather. Click a fuel type in the legend below the bar for details.
🔋 Grid Fundamentals — Key Concepts
Expandable cards covering core grid concepts — click any card to reveal definition and context. Concepts include: Frequency (60 Hz), Balancing Authority, Duck Curve, Capacity Factor, LMP (Locational Marginal Price), Congestion, Reserve Margin, and Demand Response. Formulas are shown in monospace for use in papers.
🗺 ISO Regions Panel
Click any ISO card to see: current load (GW), peak demand, generation mix %, renewable penetration, and an explanation of that region's market structure. Color = reliability status (green/amber/red). Use this to quickly orient on a region before diving into Utilization or Queue sections.
🎛 Scenario Modeling
Interactive sliders for Solar, Wind, Storage, and Demand Response. Move a slider and watch how the generation stack and reliability metrics respond. Useful for building intuition on trade-offs — e.g., "what happens to system cost when you add 20% more solar but no storage?" The model is simplified for education; for rigorous analysis use the Utilization Analysis tab.
📚 Platform Guide Accordion
Switching to Platform Guide mode replaces the modules with a clickable accordion organized by tool. Each section covers: The big idea (what the data shows and why it matters), Words you will see (defined terms), Key numbers (sourced benchmarks), and → Next step (which tab to visit next). Designed as a self-contained onboarding guide for new researchers.
Source: EIA-930 Hourly Grid Monitor (live feed for generation mix) · LBNL Queued Up 2026 · Brattle Group (Mar 2026) · FERC Orders 2023 & Dec 2025 · ISO operational data
US Generation Mix
EIA-930
Loading EIA data...
Key concept
The US grid runs on a mix of fuels that changes hourly. Natural gas dominates (cheap, ramps fast). Nuclear provides steady baseload 24/7. Renewables are growing but depend on weather.
Grid Fundamentals
ISO Regions
Selected region
Click a region to see detailed analysis
Scenario Modeling
Solar
6%
Wind
12%
Nuclear
18%
Scenario Impact
36%
Clean Energy
385
kg CO₂/MWh
2%
Curtailment
15 GW
Evening Ramp
What this means
Adjust sliders to see how generation mix affects emissions, curtailment, and ramping.
Net Load Curve ("Duck Curve")
Understanding the duck curve
California's "duck curve" shows net load (demand minus solar/wind). Midday solar creates a deep belly. Then demand peaks at sunset while solar disappears—requiring a steep 3-hour ramp of 15+ GW.
Data Center Load Growth
| Metric | Q1 2026 | 2030 est. |
|---|---|---|
| US DC load | ~30 GW | 55-70 GW |
| % of US demand | 6.2% | 10-13% |
| PJM DC queue share | 94% | ~95% |
| ERCOT large-load queue | 238 GW | 280+ GW est. |
| Community opposition projects | ~40 | Accelerating |
Research context
Data centers dominate interconnection queues: 94% of PJM forecast load growth, 70%+ of ERCOT requests. Community opposition projects increased 6× between 2023–2025. PJM capacity auction costs rose from $2.2B (2023–24) to $16.1B (2026–27) in three auction cycles — driven 82% by DC load growth (PJM IMM, Nov 2025).
System Utilization and Rate Impact
~50%
Avg Gen Utilization
+1.4%
Rate: Status Quo
-3.4%
Rate: With DERs
$110-170B
10-yr Savings
Key insight
Half the US grid sits idle most of the year because it is built for rare peak demand hours. Adding new load where spare capacity exists spreads fixed costs across more kilowatt hours, pushing rates down. The Brattle Group modeled a 4.8 percentage point rate improvement when distributed resources create headroom versus building new centralized infrastructure.
Distributed Capacity (BYODC)
Gen capacity credit
90%
Tx capacity credit
25%
Dist capacity credit
40%
DER deploy time
1-3 yrs
New gen deploy time
5-10 yrs
The concept
Data centers fund heat pumps, batteries, and weatherization in nearby homes. This reduces local peak demand, creates grid headroom, and lets the data center connect faster. Households get lower bills and better resilience. The grid avoids years of new transmission construction.
Flexibility Policies and Grid Enhancing Technologies
Active Flexibility Policies
TX SB 6 — Large loads (75+ MW) must curtail during grid emergencies. First enacted state mandate.
PJM BYOG — Expedited interconnection for loads that bring their own generation. FERC approved Jan 2026.
CA CPUC R.24-01-018 — Active rulemaking exploring BYOC, flexibility, and cost allocation for large loads.
IL HB 3437 — Would require data centers to bring clean capacity equal to max demand, broadly defined to include DERs.
Grid Enhancing Technologies
Dynamic Line Ratings — Use real-time weather to increase safe power flow. PPL: +15-17% capacity. Required by FERC Order 881.
Advanced Conductors — High-temperature, low-sag conductors carry more current on existing towers.
Power Flow Controllers — Redirect electricity around congested paths. WECC TCSC: +800 MW intertie.
DSX Flex — NVIDIA platform enabling data centers to flex 25% of load for ~200 hrs/yr. 6 major energy partners.
Why this matters
Regulators increasingly view large loads as potential grid assets. Flexibility commitments can accelerate interconnection, reduce ratepayer burden, and defer expensive infrastructure. The question is shifting from "can data centers get power" to "can they make the grid work better for everyone."
Interconnection Queue Crisis
US active queue (Q1 2026)
~1,420 GW
Active projects
~9,800
Historic completion rate
13%
PJM IC cost (active, 2020–22)
$240/kW
PJM IC cost (AI factory withdrawn)
$700–900/kW
Context (Q1 2026)
The interconnection queue is the single largest bottleneck to clean energy deployment and data center growth. Network upgrade costs drive >80% of cost increases (LBNL 2026). FERC Order 2023 cluster reforms processing first batches — AI factory projects face 8.5+ year average timelines. MISO LRTP Tranche 1.2 ($9.1B) approved Feb 2026 to begin addressing Midwest constraints.
Retail Electricity Rates by State
EIA-861M
ISO-NE
~25¢
Highest Region
BPA
~11¢
Lowest Region
2.3×
High-to-Low Spread
50
States Tracked
36+
Months of History
Key insight
Wholesale prices (LMPs) are what generators receive. Retail rates are what households pay — and include generation, transmission, distribution, and policy charges. The gap between wholesale and retail prices reveals how much infrastructure cost, regulation, and policy decisions add to electricity bills. Regions with high wholesale-to-retail spreads often have the most room for distributed capacity to reduce costs.