Hyper-Scale Energy Intelligence Platform

Where AI's energy
appetite meets
sovereign grid reality.

Practitioner-grade analysis of 14.9 GW in proposed AI data center capacity across 11 clusters in the US, China, and India—intersected against live grid constraints, renewable energy access, water stress, carbon intensity, policy risk, and infrastructure investment sizing.

14.9 GW
Tracked capacity pipeline
11
Cluster sites analyzed
3
Sovereign grids
$4.2T
Infrastructure investment implied
By Prakash Krishnamachari  ·  Digital, Data & Innovation Leader  ·  25+ years in asset-heavy infrastructure
The Strategic Context

This isn't a power story.
It's a sovereignty story.

Every hyperscaler's site selection decision is fundamentally a bet on three sovereignties simultaneously: energy sovereignty (can the grid supply and sustain the load?), regulatory sovereignty (will policy allow it to scale?), and water sovereignty (is there cooling capacity without ecological debt?). The industry tends to obsess over renewable percentages. Operators who have actually run infrastructure at scale know the real constraint is interconnect queue depth and grid inertia.

⚡
Grid Inertia Is the Hidden Constraint
Nameplate renewable capacity means nothing at the substation. The PJM interconnect queue in the mid-Atlantic region has grown to well over an order of magnitude beyond realistic near-term absorption capacity. Operators building there are essentially financing grid upgrades for the entire region.
Queue-to-absorption ratio: analyst estimate — well over 10:1 in PJM (see FERC Order 2023 backlog data) / ~3:1 estimated in ERCOT · Practitioner estimate
💧
Water Is the Silent Dealbreaker
A 100 MW data center using evaporative cooling consumes approximately 500,000 gallons of water per day (Lawrence Berkeley National Lab, 2023 estimate). Phoenix and Inner Mongolia both offer cheap power, but face severe long-run water stress that will likely trigger regulatory intervention within a 10-year asset depreciation horizon.
Water stress tier (WRI Aqueduct 4.0): Phoenix Extremely High  |  Navi Mumbai Medium–High · Source: WRI
🌿
Renewable % Is a Marketing Number
24/7 carbon-free energy (CFE) matching is the actual sustainability benchmark—not annual renewable percentage claims. Only a handful of sites globally can credibly commit to hourly CFE matching at hyperscale. Guizhou's hydro profile is the closest analog in emerging markets.
Modelled hourly CFE score (0–100): Guizhou ~90 / Inner Mongolia ~60 / Rajasthan ~55 · Modelled — see Methodology
🏛️
Policy Risk Has a Compounding Half-Life
Data center permitting cycles in the US average 3–5 years (Jones Lang LaSalle, 2024). India's wheeling regulations change faster than most hyperscalers can lock in PPAs. China's data localization and cross-border data flow restrictions add a geopolitical premium to otherwise compelling energy economics.
Avg US permitting timeline: ~36–48 months (JLL estimate)  |  India PPA lock-in risk: High · Practitioner estimate
Geospatial Intelligence Layer

11 Clusters. One Constraint Matrix.

Each site is sized by proposed MW capacity. Color indicates renewable access tier. Click any cluster for full grid intelligence including PUE targets, water stress, carbon intensity, and operator landscape.

Very High Renewable Access
High Renewable Access
Moderate Renewable Access
Low / Constrained Access
Circle size = proposed MW capacity
Multi-Dimensional Site Score

Comparative Siting Matrix

Scored across six dimensions, each weighted equally at 1/6 for a composite out of 100. All dimension scores are modelled indicators synthesized from the public datasets listed in the Methodology section — not direct utility measurements. See the Data Dictionary below for exact construction logic.

Site Proposed MW ① Renewable Access ② Grid Stability ③ Water Risk ④ Policy Clarity ⑤ Cost Index ⑥ Composite Score ⑦ Verdict
Data Dictionary — Score Construction
① Proposed MW
Pipeline capacity drawn from operator announcements, national energy authority filings, and data center industry trackers (DCD, Structure Research). Reflects announced pipeline, not commissioned capacity.
② Renewable Access Score (0–100) · Modelled
Composite of: (a) grid renewable generation share (IEA 2024 national/provincial statistics), (b) availability of open-access or wheeling mechanisms (CEA, FERC, NEA), and (c) proximity to utility-scale renewable assets (GEE spatial layer). Equally weighted thirds.
③ Grid Stability Score (0–100) · Modelled
Derived from: (a) SAIDI/SAIFI reliability indices where published (NERC, POSOCO, NEA), (b) interconnect queue depth relative to regional absorption capacity (FERC queue data), and (c) dual-substation or redundant feed availability. Qualitatively adjusted for market structure (ERCOT island risk, China state dispatch).
④ Water Risk Score (0–100) · Modelled — Higher = Lower Risk
Inverted WRI Aqueduct 4.0 Baseline Water Stress score for the relevant watershed, scaled 0–100. Score of 100 = negligible stress; score of 0 = extremely high stress. Phoenix (~8) and Inner Mongolia (~12) score critically low; Northern Virginia (~78) and Guizhou (~80) score well.
⑤ Policy Clarity Score (0–100) · Modelled
Practitioner scoring across: (a) permitting timeline predictability (JLL, Cushman & Wakefield DC market reports), (b) data sovereignty and cross-border data flow constraints (CAC, DPDP Act, CLOUD Act framework), (c) PPA/wheeling regulatory stability (CEA, FERC orders history), and (d) geopolitical risk for non-domestic operators.
⑥ Cost Index (0–100) · Modelled
All-in power cost estimate (BloombergNEF PPA tracker, Lawrence Berkeley Lab LBNL-2001975, CERC/ERCOT tariff data) normalized to 0–100 scale. Score of 100 = lowest cost in dataset (Rajasthan ~₹1.8–2.4/kWh); score of 0 = highest cost. Does not include land, construction, or labour cost differentials.
⑦ Composite Score (0–100) · Modelled
Simple unweighted average of scores ②–⑥ plus Infrastructure Readiness (operator density, substation availability, fibre maturity — drawn from Uptime Institute Global Survey 2024 and operator disclosures). All six dimensions weighted equally at 1/6. This is an analytical heuristic, not an optimisation model.
Carbon Intensity (gCO₂/kWh) · Modelled
Annual average grid carbon intensity from electricityMaps (where available) and IEA Electricity Information 2024, at the provincial/state level. Reflects location-based Scope 2 intensity. Not hourly-matched; actual operational CFE scores will differ. Displayed in map popups only.
All scores represent the author's analytical synthesis as of May 2026. They are directional indicators for comparative site evaluation and should not be used as the sole basis for investment, planning, or procurement decisions. Underlying data sources are cited in the References & Methodology section.
Regional Intelligence

Inside the Grid Rooms

Per-site operational intelligence including operator landscape, power purchase agreement structure, cooling architecture constraints, and the unfiltered grid constraint picture.

Northern Virginia
Loudoun Co. / PJM Grid
3,200
MW PROPOSED
Grid Constraint
Severe PJM transmission congestion. Dominion Energy's grid operating near capacity in Loudoun County. New interconnect applications face 5–7 year timelines. "NOVA2" and "NOVA3" substations both at saturation.
Renewable Pathway
Offshore wind (CVOW, 2.6GW) is the primary incremental source. Virginia's 100% clean energy mandate by 2045 creates policy tailwind but execution risk. PPA rates: ~$45–55/MWh for wind.
Key Operators
Amazon (AWS), Microsoft, Google, Meta, CyrusOne, Equinix. Combined land holdings in Loudoun exceed 2,500 acres.
PUE Target Range
1.10–1.15
Excellent
Water Stress Index
Low (Potomac watershed)
Phoenix Metro
Maricopa Co. / APS + SRP Grid
1,400
MW PROPOSED
Grid Constraint
Stable desert distribution grid. WECC interconnect with favorable capacity margins. High solar alignment—load peaks during daylight hours when generation peaks. Transmission headroom exists in Palo Verde corridor.
Renewable Pathway
Solar irradiance among highest in North America (6.5 kWh/m²/day). Utility-scale solar PPA rates: ~$25–35/MWh. Agnosec/APS 2032 renewable portfolio at 65%. Battery storage co-location becoming standard.
Key Operators
Microsoft (Phoenix-Goodyear campus, 20-site expansion), Meta, Google, Switch, CyrusOne.
PUE Target Range
1.25–1.35
Challenged by heat
Water Stress Index
Extremely High
Dallas–Fort Worth
Allen/Plano / ERCOT Grid
1,800
MW PROPOSED
Grid Constraint
ERCOT is an island grid—no federal FERC oversight, enabling faster PPAs. But 2021 Winter Storm Uri exposed catastrophic single-point failure risk. New hardening rules exist but long-run reliability remains contested.
Renewable Pathway
Texas is the #1 wind state (38 GW installed). Negative pricing events create exceptional economics for flexible loads. Solar buildout accelerating in West Texas with new CREZ transmission lines.
Key Operators
Oracle, Nokia, T5, Stream, Flexential, Iron Mountain. State tax incentives (Chapter 313 successors) remain potent.
PUE Target Range
1.12–1.20
Good
Water Stress Index
Medium-High
Guizhou Hub
Guiyang / National Policy Anchor
1,200
MW PROPOSED
Grid Context
Part of China's "东数西算" (East Data, West Computing) national strategy. Guizhou designated as one of 8 national data center clusters. Hydro-dominant grid (Wujiang River system) provides near-baseload clean power.
Renewable Profile
Hydroelectric dominance (78% of provincial generation mix). Year-round mild temperatures (annual average 15°C) enable natural cooling. Lowest carbon intensity grid in the analysis at ~82 gCO₂/kWh.
Key Operators
Alibaba Cloud (Guizhou Data Center Park), Tencent, Huawei Cloud, Apple iCloud China. State Grid designated backbone.
PUE Achieved
1.08–1.12
Best-in-class
Geopolitical Risk
High for non-Chinese operators
Inner Mongolia
Ulanqab / Northern Grid Corridor
950
MW PROPOSED
Grid Context
Direct connection to Northern China UHV power export corridor. Power price among China's lowest at ~¥0.25/kWh industrial rate. Winter free-cooling season: October through April (ambient -20°C to +5°C).
Renewable Profile
Wind-dominant with some solar. Curtailment rates historically 15–20% (improving). Carbon intensity: ~420 gCO₂/kWh (still coal-heavy in winter). Green certificate market emerging.
Key Operators
Alibaba, Baidu, China Mobile, China Telecom. Government-subsidized land and power pricing through 2030.
PUE Achieved
1.09–1.14
Excellent (free cooling)
Water Stress Index
Very High (arid steppe)
Yangtze River Delta
Shanghai / Eastern Grid
1,600
MW PROPOSED
Grid Context
Ultra-high baseline industrial demand creates constant upward pressure on available capacity. Grid density is extreme; new large-load connections face 3–4 year queues in priority zones.
Renewable Profile
Low renewable fraction locally (30%). Relies on UHV west-to-east power transfer. High cost premium. Primary value: latency to Shanghai financial district and subsea cable connectivity.
Key Operators
GDS Holdings, 21Vianet, China Telecom IDC, Equinix (via partnerships). Financial services latency requirement drives siting premium.
Carbon Intensity
~550 gCO₂/kWh
Mumbai / Navi Mumbai
Maharashtra / MSEDCL Grid
2,200
MW PROPOSED
Grid Context
Highest grid reliability in India by SAIDI/SAIFI metrics. Multiple submarine cable landings (FALCON, SMW-4, EIG, 2Africa). CIDCO-designated MIDC zones in Navi Mumbai purpose-built for data center density.
Renewable Pathway
Maharashtra's 35 GW renewable target by 2030. Wheeling from Rajasthan/Gujarat solar farms at ~₹0.5/unit wheeling charge. Open access regulations improving under CEA 2023 amendments. PPA range: ₹2.8–3.5/kWh solar.
Key Operators
CtrlS, Yotta, NTT, Equinix, ST Telemedia GDC, Web Werks, Digital Realty. 850+ MW under construction in Navi Mumbai alone (2024–26).
PUE Target Range
1.45–1.65
Cooling-challenged climate
Water Stress Index
Medium (monsoon-dependent)
Bengaluru
Karnataka / BESCOM Grid
850
MW PROPOSED
Grid Context
Stable technology park distribution via BESCOM. Karnataka's progressive wheeling and open-access laws (Electricity Regulatory Commission orders 2022-23) have made renewable procurement significantly more feasible than rest of India.
Renewable Pathway
Altitude (920m elevation) enables adiabatic cooling with minimal energy overhead. Karnataka wind-solar hybrid zones in Chitradurga/Davangere accessible via open access. Effective renewable procurement cost: ₹4.2–5.1/kWh all-in.
Key Operators
Google (Hyderabad/Bengaluru hybrid), Microsoft, IBM, CTRL-S, Netmagic, AWS (edge nodes). Talent proximity premium for AI/ML workloads is a key differentiator.
PUE Achieved
1.40–1.55
Better than Mumbai
Water Stress Index
Medium-High (Cauvery basin tension)
Rajasthan Solar Zone
Jaisalmer / RVPN Grid
600
MW PROPOSED
Grid Context
Emerging edge: RVPN grid expanding rapidly to absorb 30+ GW in committed renewable projects. Transmission infrastructure is the primary constraint—new 765kV lines scheduled for commissioning through 2027.
Renewable Profile
India's highest solar irradiance zone (6.8–7.0 kWh/m²/day). Wind resource also excellent in Jaisalmer corridor. First-mover advantage for hyperscale campuses co-located with generation assets. PPA potential: ₹1.8–2.4/kWh.
Key Operators
Greenfield—no hyperscale operator has commissioned here yet. Adani Green and ReNew Power hold land positions adjacent to potential data center zones.
Water Stress Index
Extremely High — air-cooling mandatory
Data Analytics Layer

Visualizing the Constraint Landscape

Four lenses that practitioners need: power cost versus renewable access, carbon intensity versus proposed scale, PUE performance versus climate zone, and the regional capacity pipeline breakdown.

Renewable Access Score vs Power Cost
Identifying sites where clean energy and economics align
Proposed MW Capacity by Region
US leads pipeline volume; China leads per-cluster density
Carbon Intensity vs Scale (gCO₂/kWh)
The sustainability gap between nameplate and reality
Composite Site Score Ranking
Multi-dimensional siting score (0–100)
Infrastructure Architecture Principles

What "Renewable-Ready" Actually Requires

Moving from renewable energy commitment to operational 24/7 carbon-free infrastructure requires architectural choices at six layers—most of which are irreversible once site construction begins.

01
Grid Interconnect Architecture
The physical point of connection to the utility determines your entire renewable flexibility envelope. Dual-substation designs with separate feeder paths add 15–20% capex but halve outage risk. For sites exceeding 200 MW, private substation ownership is non-negotiable.
132/220kV Dual Feed Grid Forming Inverters BESS Buffer
02
Cooling System Architecture
ASHRAE A2 (35°C intake) is the floor. AI/GPU workloads running at 60–100kW per rack demand direct liquid cooling (DLC) as the baseline—not air cooling with DLC supplement. Water-side economizer design must account for wet bulb temperature exceedance hours over 30-year site life.
Direct Liquid Cooling Rear-Door Heat Exchangers Adiabatic Pre-Cooling
03
Energy Storage & Buffering
Renewable intermittency at hyperscale requires 4–8 hours of behind-the-meter battery storage to achieve smooth load profiles. Lithium iron phosphate (LFP) chemistry at this scale now pencils in at $180–220/kWh installed, dropping 12% annually. Flow batteries emerging for longer-duration buffering.
LFP BESS Vanadium Flow Demand Flexibility
04
Power Delivery & Conversion
Modern hyperscale power chains: utility → HV switchgear → medium-voltage distribution → UPS → server PSU → VRM. Every conversion step loses 2–4%. Moving to 48V DC distribution at rack level, eliminating one conversion stage, can recover 150–300 kW per MW of IT load.
48V DC Distribution Li-Ion UPS N+1 Generator Topology
05
24/7 CFE Procurement Structure
Annual renewable energy certificates (RECs) are the legacy instrument. Hourly Power Purchase Agreements (H-PPAs) are the emerging standard—matching consumption to generation on an hourly basis. This requires sophisticated load-shifting capability and energy management systems that most operator-class facilities don't yet have.
Hourly PPA Virtual PPA Additionality Matching
06
Operational Intelligence Layer
Carbon-aware computing—shifting workloads to hours and locations where the grid is cleanest—can reduce effective carbon intensity by 20–40% without hardware changes. Requires real-time grid carbon intensity signals, flexible orchestration middleware, and workload classification systems to identify deferrable versus latency-sensitive jobs.
Carbon-Aware Scheduling Real-Time Grid Signals Workload Orchestration
Structural Risk Register

The Risks Operators Don't Publish

Beyond site selection scores, these are the systemic risks that will reshape the landscape over the next decade—drawn from infrastructure deployment experience across Shell, Maersk, and energy sector operations.

HIGH PROBABILITY · HIGH IMPACT
Interconnect Queue Collapse
The US interconnect study process is structurally broken. FERC Order 2023 reforms will improve throughput, but the existing 1,000+ GW backlog means greenfield sites committing power today face realistic 6–9 year wait times. Operators anchoring 2027 buildout timelines to new interconnects are carrying undisclosed schedule risk.
HIGH PROBABILITY · HIGH IMPACT
Water Regulation Acceleration
Arizona, California, and Maharashtra are all moving toward mandated water efficiency standards for large industrial consumers. A 500 MW air-cooled campus in Phoenix consumes approximately 1.5 billion gallons annually. Facilities designed today to current standards face forced retrofits within a 5-year window as regulation catches up to climate science.
MEDIUM PROBABILITY · HIGH IMPACT
India Open Access Regulatory Reversal
State electricity boards in India have a documented history of restricting open access wheeling when it threatens distribution company revenue. Maharashtra, Telangana, and Karnataka have all imposed temporary restrictions within the last 5 years. PPAs signed today carry material volume risk if regulatory access is curtailed.
MEDIUM PROBABILITY · MEDIUM IMPACT
China Data Localization Escalation
The 2021 Data Security Law and Personal Information Protection Law created a new compliance layer. MLPS 2.0 and CAC oversight rules are tightening. Non-Chinese operators face an increasing compliance cost burden that erodes the economic advantage of Guizhou's energy and PUE economics.
MEDIUM PROBABILITY · HIGH IMPACT
Grid Carbon Intensity Disclosure Mandates
SEC climate disclosure rules (Rule S-K) and EU CSRD both mandate Scope 2 location-based and market-based reporting. Sites in coal-heavy grids (Inner Mongolia, Yangtze Delta) will carry a disclosure liability that currently isn't priced into land and power cost comparisons.
LOW PROBABILITY · HIGH IMPACT
Hydrogen Economy Disruption
If green hydrogen achieves $2/kg at scale (currently $5–8/kg), the entire cooling and backup power architecture for data centers shifts. Fuel cell-based backup eliminates diesel generator requirements and transforms the stranded asset risk profile of sites currently dependent on fossil backup for N+1 redundancy.
Practitioner's Synthesis
"The hyperscale energy conversation is dominated by press releases and policy announcements. The actual constraint is always in the last 10 meters—the transformer, the cooling tower, the interconnect agreement."
Guizhou's hydro story is genuinely compelling, but the data sovereignty constraints make it operationally inaccessible to most Western operators. Northern Virginia's operator density is a competitive moat, but also a grid saturation trap. India's solar economics are extraordinary, but the wheeling regulatory environment requires active management, not passive assumption.

The most underappreciated opportunity in this analysis is the Rajasthan Solar Zone—first-mover economics for an operator willing to own their generation assets and accept 2027–2028 infrastructure maturity. The sites that win the next decade are those where the operator is willing to function as a quasi-utility, not just a power consumer.
Technical Architecture

From GEE Pipeline to Insight

This platform was seeded by a Python spatial intelligence engine (generate_dashboard.py) built on Google Earth Engine (GEE) data primitives and Folium map rendering. The v2 upgrade extended that foundation with multi-source data ingestion, a six-dimensional scoring model, and a self-contained single-file deployment architecture.

Spatial Intelligence Pipeline — Data Flow
Layer 1
INGESTION
REST API / Open-Data Pipeline
US: FERC interconnect queue exports, EIA-860 generator data, NERC reliability reports · China: NEA provincial generation statistics, State Grid capacity disclosures · India: CEA annual reports, POSOCO SAIDI/SAIFI tables, WRIS water resource data · Cross-cutting: WRI Aqueduct 4.0 GeoJSON, electricityMaps carbon intensity API, BloombergNEF PPA tracker extracts
Layer 2
GEE SPATIAL
Google Earth Engine Spatial Processing
GEE JavaScript / Python API used to extract: solar irradiance rasters (NASA POWER / PVGIS integration), wind capacity factor grids (Global Wind Atlas), water stress polygons (WRI Aqueduct), and land-use suitability layers (ESA WorldCover 2021). Coordinate translation matrices applied via GeoPandas to align multi-CRS datasets to WGS84 decimal degrees for Leaflet rendering.
Layer 3
SCORING
Six-Dimension Scoring Model
Each of the six dimensions (renewable access, grid stability, water risk, policy clarity, cost index, infrastructure readiness) is constructed from 2–4 normalised sub-indicators, then averaged. Normalisation uses min-max scaling across the 11-site dataset. Composite score = unweighted mean of six dimension scores. All scores are analytical estimates — not utility measurements.
Layer 4
RENDERING
Folium Map Core → Leaflet.js · Chart.js
Original pipeline: Python Folium engine generating CircleMarker layers with radius ∝ √MW capacity, colour-classified by renewable tier (green → amber → red), compiled to Base64 data-stream and embedded in iframe. v2 upgrade: ported to vanilla Leaflet.js for direct DOM control, added Chart.js analytics layer (scatter, bar, bubble charts), and replaced iframe embed with native map div for full interactivity.
Layer 5
DEPLOYMENT
Self-Contained Single-File Architecture → GitHub Pages
Output: single index.html (~2,100 lines) with all CSS, JavaScript, and data embedded inline. No build toolchain, no npm, no server required. All external dependencies (Leaflet, Chart.js, Google Fonts) loaded from CDN. Deployable to GitHub Pages by replacing one file. Original Python pipeline preserved as generate_dashboard.py in repository root.
🌍
Google Earth Engine
Spatial raster extraction: solar irradiance, wind capacity factors, water stress polygons, land cover classification.
🐍
Python / Folium
Seed pipeline: GeoPandas alignment, coordinate translation, Folium CircleMarker generation, Base64 compilation to standalone HTML.
🗺️
Leaflet.js + Chart.js
Interactive map (CartoDB Positron tile), proportional circle markers, rich popups, region filter controls, and four analytics chart panels.
🚀
GitHub Pages
Zero-infrastructure deployment. Single HTML file pushed to repository root. Live at prakashkrish-datageek.github.io/Ecogrid-AI.
References & Methodology

Data Sources, Limitations & Disclosure

This analysis synthesises public datasets, industry reports, and practitioner judgement. All modelled scores are directional indicators for comparative evaluation — not utility measurements or investment-grade assessments. Each source is listed with the specific data element it informs.

Primary Data Sources
⚡ Grid & Interconnect
  • →FERC Interconnection Queue (ferc.gov) — US interconnect backlog volumes and study timelines. Informs Grid Stability Score and PJM queue context.
  • →NERC Long-Term Reliability Assessment 2024 — US regional reserve margins and SAIDI/SAIFI reliability metrics.
  • →POSOCO Annual Report 2023–24 (India) — State-level distribution reliability indices. Informs India Grid Stability scores.
  • →China NEA Provincial Generation Statistics 2024 — Provincial generation mix and renewable share. Informs China Renewable Access scores.
  • →CEA Annual Report 2023–24 (India) — State-level renewable capacity and open access framework data.
💧 Water & Environment
  • →WRI Aqueduct 4.0 (wri.org/aqueduct) — Baseline Water Stress scores at watershed level. Directly feeds Water Risk Score (inverted and scaled).
  • →Lawrence Berkeley National Lab — United States Data Center Energy Usage Report (LBNL-2001975) — Water consumption per MW benchmarks for cooling systems.
  • →ESA WorldCover 2021 (via GEE) — Land cover classification for siting constraint analysis.
🌿 Renewable Energy
  • →IEA World Energy Statistics 2024 / Electricity Information 2024 — National and provincial generation mix. Informs carbon intensity estimates and Renewable Access scores.
  • →electricityMaps (electricitymaps.com) — Real-time and historical grid carbon intensity at country/zone level. Carbon intensity values in map popups are annual averages derived from this source.
  • →Global Wind Atlas 3.0 (DTU / World Bank, via GEE) — Wind capacity factor rasters at 250m resolution. Informs renewable access scoring for wind-dominant sites (Inner Mongolia, DFW, Rajasthan).
  • →NASA POWER / PVGIS (via GEE) — Solar irradiance (GHI) rasters at 1km resolution. Informs renewable access scoring for solar-dominant sites (Phoenix, Rajasthan, Inner Mongolia).
Industry Reports & Practitioner Sources
🏗️ Data Center & Infrastructure
  • →Uptime Institute Global Data Center Survey 2024 — PUE benchmarks by region and cooling architecture type. PUE target ranges in site profiles are calibrated against this data.
  • →BloombergNEF New Energy Outlook & PPA Tracker 2024 — Power purchase agreement pricing by technology and geography. Feeds Cost Index and per-site PPA rate estimates.
  • →JLL Data Center Outlook 2024 / Cushman & Wakefield Global DC Report — Permitting timelines, operator density, land availability. Feeds Policy Clarity and Infrastructure Readiness scores.
  • →Data Center Dynamics (DCD) / Structure Research — Pipeline MW tracking, operator announced projects, commissioning timelines. Primary source for Proposed MW figures.
  • →The Green Grid — PUE Metrics & Best Practices — PUE definition, benchmarking methodology, and cooling architecture classification used in site profiles.
⚖️ Policy & Regulation
  • →FERC Order 2023 — US interconnect queue reform framework. Context for PJM backlog characterisation.
  • →China CAC Data Security Law (2021) / PIPL — Data sovereignty constraints applied in China Policy Clarity scoring.
  • →India CERC Open Access Regulations / DPDP Act 2023 — Open access wheeling rules and data localisation obligations. Informs India Policy Clarity scores.
⚠ Limitations & Disclosure
  • ·All dimension scores are modelled estimates, not direct utility measurements. They reflect the author's analytical synthesis and should be treated as directional indicators only.
  • ·Proposed MW figures reflect announced pipelines, not commissioned or under-construction capacity. Actual build-out will differ materially.
  • ·Carbon intensity values are annual provincial averages. Hourly CFE scores are modelled approximations based on generation mix and weather patterns — not metered data.
  • ·This analysis does not constitute investment, legal, or engineering advice. Data compiled May 2026; grid and policy environments change rapidly.
  • ·China data is derived from state-published statistics; independent verification is limited.