Utilities generate, transmit and distribute electricity, and deliver natural gas and water, mostly under state or federal rate regulation. Most earn an allowed return on the infrastructure they build.
Representative public companies: NextEra Energy (NEE), Southern Company (SO), Duke Energy (DUK), American Electric Power (AEP), Dominion Energy (D), Constellation Energy (CEG), Vistra (VST), Entergy (ETR). Named for context only, not as recommendations.
Research current as of September 12, 2026 · Adoption stage: Bifurcated
For utilities, AI shows up mostly as demand, while AI inside operations is still uneven. AEP said on July 30, 2026 that it has 69 gigawatts of load additions contracted through 2030, and Southern Company said data center electricity use rose 55% year over year in Q2 2026. Internal use is narrower: PG&E uses AI to detect smoke on more than 630 cameras, and NextEra announced in December 2025 a plan to use Google's forecasting and agentic AI in grid and field operations. Google has built 1 gigawatt of demand response into utility contracts, cutting machine learning workloads when grids are stressed. The biggest open question is reliability: on July 16, 2026, FERC ordered NERC to file mandatory standards for computational loads by December 31, 2026, without setting a size threshold.
| Company | What was deployed | Date |
|---|---|---|
| PG&E (PG&E Corporation, PCG) | AI smoke detection on high-definition wildfire cameras and machine learning weather models to predict outages and fire risk More than 630 cameras monitor 90% of high fire-risk areas in its service territory |
2025-11-07[1] |
| NextEra Energy | Partnership with Google Cloud to apply TimesFM 2.5 and WeatherNext 2 forecasting and agentic AI to grid planning, equipment issue prediction and crew scheduling, plus joint development of gigawatt-scale data center campuses | 2025-12-08[2] |
| Google with Indiana Michigan Power (AEP), TVA, Entergy Arkansas, Minnesota Power and DTE Energy | Demand response written into long-term energy contracts, under which Google limits or shifts machine learning workloads when utilities need to reduce load 1 GW of data center demand response capacity contracted |
2026-03-19[3] |
| American Electric Power | Contracted large-load additions, mostly from hyperscalers and large industrial customers, backed by large-load tariffs meant to shield existing customers 69 GW of additions contracted through 2030, including 6 GW added in Q2 2026; five large-load tariffs approved and three pending |
2026-07-30[4] |
| Southern Company | Connection of data center load and contracting of large-load agreements across its utilities Data center electricity use up 55% year over year in Q2 2026; more than 1.2 GW of operating data center load, up more than 500 MW in a year; more than 17 GW contracted by the mid-2030s |
2026-08-18[5] |
| Dominion Energy | Contracting of data center capacity in Virginia under a large-load framework that requires those customers to fund their own infrastructure More than 53 GW of data center capacity in various contracting stages, about 12 GW under electric service agreements |
2026-07-31[6] |
Projections grounded in what is happening now, not certainties. Each one names the signal that would confirm it.
Utilities face a two-sided squeeze: data center demand is enormous (bullish) but hyperscalers building off-grid power is a direct moat-breaker (bearish). Meta captive nuclear, Amazon/Google same — 33% of data centers could be off-grid by 2030. Natural monopoly + rate-of-return regulation still protects the core business, but the highest-value new customers (hyperscalers) are bypassing utilities entirely.
AI optimizes grid operations internally but the growth thesis depends on whether utilities capture data center demand or lose it to captive generation. compDyn flipped negative in v4 for this reason. BAUMOL NOTE: Physical infrastructure maintenance and grid operations remain stubbornly human-intensive.
As cognitive sectors surge in AI productivity, utility field labor costs get pulled upward by economy-wide wage competition — while rate-of-return regulation limits utilities' ability to pass costs through quickly.
A score of 1.00 = no change.
>1.00 means AI (net) boosts this industry's relative value.
<1.00 means AI (net) erodes it.
These are relative scores — how this industry's share of the economy shifts in relation to all 31 industries we track.
| Timeframe | Score | Range | In Plain English |
|---|---|---|---|
| 1 Year | 1.14 | 1.12 – 1.16 | Slight tailwind (+14%) |
| 2 Years | 1.24 | 1.19 – 1.28 | AI tailwind (+24%) |
| 3 Years | 1.50 | 1.42 – 1.58 | AI tailwind (+50%) |
| 5 Years | 2.07 | 1.92 – 2.22 | AI tailwind (+107%) |
| 10 Years | 2.44 | 2.24 – 2.64 | AI tailwind (+144%) |
The same engine run under three sets of assumptions. The spread between the rows shows the uncertainty. It is not a forecast of which one happens.
| Scenario | 1 yr | 2 yr | 3 yr | 5 yr | 10 yr | Rank |
|---|---|---|---|---|---|---|
| Boom Faster AI progress, energy breakthroughs | 1.14 | 1.23 | 1.51 | 2.17 | 2.62 | #22 |
| Base Base case AI progress, current trajectory energy | 1.14 | 1.24 | 1.50 | 2.07 | 2.44 | #20 |
| Doom Slower AI progress, constrained energy, market correction | 1.10 | 1.22 | 1.47 | 1.96 | 2.28 | #13 |
Each component is scored from research, weighted differently at each time horizon, and added together:
Then cascaded through 205 cross-industry relationships.
Built from earnings calls, patent filings, regulatory documents, deployment disclosures and peer-reviewed research. 205 mapped cross-industry relationships. Recalibrated continuously since February 2026, with every parameter change logged.
Refreshed with every major development. Real deployment numbers, quarterly earnings data, regulatory shifts (EU AI Act, export controls), and macro conditions like the energy bottleneck. Not a static model — a living one.
This report uses: Base case AI development pace and Current trajectory energy trajectory. Members can adjust both and see how scores change in real time.
When AI changes these industries, it ripples into Utilities.
| Industry | Effect | Strength | Timing |
|---|---|---|---|
|
Energy
Energy infrastructure investment drives utility grid expansion and modernization
|
Positive | 90% | Near-term |
|
Technology Hardware & Equipment
AI growth in Technology Hardware & Equipment creates demand or opportunities that benefit Utilities
|
Positive | 85% | Near-term |
|
Real Estate
AI growth in Real Estate creates demand or opportunities that benefit Utilities
|
Positive | 80% | Medium-term |
|
SOFTWARE - Cloud & AI Platforms
AI changes in SOFTWARE - Cloud & AI Platforms have complex, mixed effects on Utilities
|
Mixed | 75% | Near-term |
|
Capital Goods
AI growth in Capital Goods creates demand or opportunities that benefit Utilities
|
Positive | 70% | Medium-term |
|
Food, Beverage & Tobacco
AI growth in Food, Beverage & Tobacco creates demand or opportunities that benefit Utilities
|
Positive | 30% | Near-term |
When AI changes Utilities, 6 other industries feel it:
See how all 31 industries compare in our Full AI Market Cascade Dashboard