AI ENERGY

US Frontier AI Data Center Power Monitor

INITIALIZING...

Read the full analysis

The Grid War: AI Has Run Out of Power

Crisis Timeline
Key milestones and projected inflection points
Dec 2, 2025
NOWLIVE

Current Status

DEFCON 4 - Severe

Current
Q4 2025

Palisades Nuclear Restart

800 MW baseload capacity returns

0
days
Q2 2026CRITICAL

The Depreciation Bomb

H100 chip obsolescence crossover

180
days
Q1 2027CRITICAL

Transformer Lead Time Apex

Peak supply chain bottleneck

453
days
2028

Grid Capacity Catch-Up

Projected equilibrium point

911
days

Critical Window: Q2 2026 - Q1 2027

Peak collision between AI deployment pressure and transformer supply bottleneck. H100 generation enters depreciation curve while H200/B200 shipments accelerate without power infrastructure.

180days to depreciation bomb
453days to transformer apex

Understanding the Deficit

Loading power deficit data...

Loading transformer queue data...

Grid Reality Layer
EIA-930 (baseline)
Reserve margins across 6 major ISOs
The Deployment Gap
Bain/McKinsey (2024)
AI Data Centers Only
AI data center power demand vs. allocated grid capacity (GW)

Note: This chart shows AI-specific demand (17 GW gap in 2025). The "Total Grid Gap" in the Famine Clock (47.2 GW) includes additional demand from EVs, industrial electrification, and reserve margin erosion.

WE ARE HERE

Year

2025

17 GW

12.8M homes

Gap Growth

+20

GW/yr

2025

17

GW

2026

37

GW

2028

67

GW

2030

78

GW

Exponential Gap Growth

AI demand growing at ~40% CAGR while AI-allocated grid capacity grows at ~8% CAGR. By 2030, the 78 GW AI deficit equals 97 nuclear reactors worth of missing power.

Stranded Silicon Monitor
GPU deployment vs shipping gapEpoch AI
Epoch AI (Nov 26, 2025)
25%
50%
75%

Stranded GPUs

2.28M

46%

of shipped

Deployed GPUs

2.72M

54%

utilized

Stranded Power

1.8 GW

1.4M homes

Idle Value

$0.0B

Active Power

4.1 GW

3.1M homes

Data Centers

9

The Bottleneck: NVIDIA shipped ~5M H100-equivalent GPUs, but only 2.7M are deployed in data centers with power. The remaining 2.3M GPUs sit idle, depreciating at 20%/year while awaiting grid capacity.

Material Bottleneck Matrix
FRED (2024)
Supply chain constraints & lead times
Energy Source Breakdown

Power sources for operational AI data centers

EIA + IEA

Aggregate Energy Mix

4,123 MW total3.1M homes
35%
25%
21%
11%
Natural Gas(35%)
Nuclear(25%)
Coal(21%)
Wind(11%)
Hydro(5%)
Solar(4%)

Fossil Fuels Dominate: 56% of AI Power

Despite corporate carbon neutral pledges, Natural Gas (35%) is the largest power source for US frontier AI data centers. The grid mix is determined by regional infrastructure, not company choice. Nuclear provides 25%, renewables (wind + solar) provide 15%.

Energy Sources by Data Center

Amazon Madison Mega Site(Amazon)
341 MW
28%
32%
22%
MISO GridGrid connected via Entergy Mississippi
Anthropic-Amazon New Carlisle(Amazon)
751 MW
28%
32%
22%
MISO GridGrid connected via NIPSCO
Google New Albany(Google Cloud)
543 MW
38%
32%
18%
PJM GridSolar PPA with TotalEnergies, grid connected
Google Omaha(Google Cloud)
189 MW
42%
32%
18%
SPP GridGrid connected via OPPD
Meta Prometheus(Meta)
615 MW
38%
32%
18%
PJM GridNuclear PPA with Constellation (Clinton plant)
Microsoft Fairwater Atlanta(Microsoft)
615 MW
45%
22%
18%
SERC GridGrid connected via Georgia Power
OpenAI Stargate Abilene(Oracle)
295 MW
42%
28%
ERCOT GridOn site gas turbines planned
xAI Colossus 1(xAI)
498 MW
42%
28%
TVA GridMobile gas turbines, TVA grid backup
xAI Colossus 2(xAI)
276 MW
42%
28%
TVA GridMobile gas turbines, TVA grid backup

Source: Regional grid mix from EIA Form 861 (2024). Energy mix reflects the regional grid where each data center is located, not direct power purchase agreements. Some facilities (xAI Memphis, Crusoe Texas) use on site gas turbines. IEA Energy and AI Report (2024) for methodology.

Power Independence Tracker

Grid reliance vs on-site generation by data center

Company filings + News

74% of AI Compute is Grid Dependent

3,054 MW of 4,123 MW total capacity relies entirely on grid infrastructure. These facilities are bottlenecked by the interconnection queue (5+ year wait). Only 667 MW has bypassed the queue via on-site generation.2.3M homes

Grid Dependent

5

2,511 MW

Grid + PPA

1

543 MW

Hybrid

2

793 MW

On-Site Primary

1

276 MW

Power Source Distribution
4,123 MW total3.1M homes
74% Grid
19%
7%
Grid (3,054 MW)
Hybrid (793 MW)
On-Site (276 MW)
Interconnection Queue Impact

3,054 MW

Fully queue bottlenecked

2.3M homes

667 MW

Bypassed via on-site generation

500K homes

Data Center Power Profiles

Anthropic-Amazon New CarlisleAmazon

Grid connected via NIPSCO

751 MW

100% grid

Meta PrometheusMeta

Grid connected via AEP Ohio

615 MW

100% grid

Microsoft Fairwater AtlantaMicrosoft

100% grid dependent via Georgia Power

615 MW

100% grid

Google New AlbanyGoogle Cloud

Nuclear PPA with Constellation (still grid delivered)

543 MW

100% grid

xAI Colossus 1xAI

Operating unpermitted turbines since 2024, TVA grid as backup

498 MW

40% grid

Mobile Gas Turbines (24+ units)150 MW on-site
Amazon Madison Mega SiteAmazon

Grid connected via Entergy Mississippi

341 MW

100% grid

OpenAI Stargate AbileneOracle

1 GW facility planning mixed on-site generation

295 MW

50% grid

Planned Gas Turbines + Solar500 MW on-site
xAI Colossus 2xAI

Scaling turbine fleet for 200k GPU expansion

276 MW

20% grid

Mobile Gas Turbines (expanded)300 MW on-site
Google OmahaGoogle Cloud

Grid connected via OPPD

189 MW

100% grid

Source: Company SEC filings, Reuters, CNBC, Shelby County Health Department permits (xAI), EIA regional grid data. On-site generation estimates based on turbine counts and manufacturer specifications.

Loading interconnection data...

What-If Simulator
Test policy interventions to mitigate the energy crisis
Epoch + LBNL

Current Power Deficit

47.2 GW

35.4M homes

Critical Threshold

Q2 2026

+0.0 GW
0 plants (Palisades, Duane Arnold, etc.)~0.9 GW each
+0.0 GW
0 SMR plants (NuScale, X-energy)~0.3 GW each
+0.0 GW
0% reduction in wait time~5 GW per 10% faster
+0.0 GW
0 additional units ordered~2 GW per 100 units
+0.0 GW
0 facilities with on-site gas turbines~0.5 GW each

Bypasses 124-week large power transformer queue. Uses readily available distribution transformers (8-16 week lead time).

Projected Outcome

CRITICAL

Original Deficit

47.2 GW

Capacity Added

+0.0 GW

New Deficit

47.2 GW

35.4M
Projected Critical Date

June 2026

No change from baseline

Methodology: Impact estimates based on DOE reactor capacity data, NuScale SMR specifications, LBNL queue analysis, and WoodMac transformer market reports. Actual results depend on execution timelines.

Frequently Asked Questions

Understanding the US AI energy infrastructure crisis

The Grid War: AI Has Run Out of Power

The unseen fight between hyperscalers, physics, and a grid built for a slower world.