For much of the industry’s history, utilities planned against gradual, distributed increases in demand. That model is being disrupted by large, concentrated loads from data center campuses, semiconductor facilities, or advanced manufacturing that can emerge faster than traditional planning and regulatory cycles.

But whether that load can be served depends on far more than available generation. Utilities must simultaneously understand transmission capacity, distribution constraints, interconnection queues, permitting timelines, equipment availability, customer flexibility, and the cost implications for existing ratepayers.

The challenge is therefore not simply forecasting demand. It is to determine where demand can be connected, how quickly it can be served, and which combination of infrastructure, procurement, and commercial structures can do so at the lowest system cost.

New infrastructure will remain essential. But permitting delays, interconnection backlogs, equipment shortages and workforce constraints mean that much of it will not arrive quickly enough to address near-term demand. The immediate opportunity is to extract more capacity, reliability and resilience from assets already in the field.

Utilities often possess large volumes of network, maintenance and inspection data, but that information remains fragmented across geographic information systems, asset registers, work-management platforms and field records. As a result, operators may know that an asset exists without having a current, reliable view of its condition, risk or available capacity.

The solution is an operational asset-intelligence layer.

For a transmission and distribution services provider, a GIS-based platform was developed to map and visualize distribution assets across the network. The platform was then integrated with an AI-powered visual inspection capability that analyzes drone-captured images and videos to automatically detect, classify, and annotate defects.

This combination matters. GIS provides the location and network context. Computer vision identifies physical deterioration. Asset data reveals maintenance history. Together, these capabilities allow utilities to prioritize interventions based on condition and operational consequences rather than on fixed inspection schedules alone.

The impact extends beyond reducing inspection effort. Better asset intelligence allows operators to identify emerging constraints earlier, target maintenance more precisely and determine where relatively small interventions could unlock additional system headroom.

The same principle applies across grid-enhancing technologies, storage optimization, dynamic line ratings, predictive maintenance and demand flexibility.

The next unit of usable capacity may not come from building a new asset. It may come from understanding the existing system well enough to operate it closer to its real limits without compromising safety or reliability.

AI is both the pressure point and part of the answer

AI appeared throughout the Reuters discussion in two seemingly contradictory roles.

It is one of the fastest-growing sources of electricity demand, driven by the expansion of large, power-dense data centres. At the same time, it is becoming one of the few technologies capable of helping utilities operate systems whose complexity is increasing faster than human planning capacity.

That does not mean utilities need more generic AI pilots. The highest-value applications are domain-specific and embedded in operational workflows: 

AI becomes useful when it is connected to the data, rules and decisions that define utility operations.

The customer domain illustrates this shift. One European utility used its existing smart-meter and data foundations to create a conversational experience that allows customers to interact directly with their energy data. Customers can ask questions about consumption, understand usage trends, and receive personalized recommendations.

The utility also implemented an AI-powered alerting capability that identifies unusual consumption patterns and notifies participating customers before they receive an unexpectedly high bill.

These are not simply customer-service features. They convert smart-meter data from a billing input into an active decision tool. Better-informed customers can respond to abnormal usage, participate in demand-management programs and engage more meaningfully with dynamic energy services.

The same principle applies on the supply side.

Renewable operators are using generation and price forecasting to support bidding decisions across day-ahead and real-time markets. By combining weather data, asset performance, historical generation and market prices, operators can improve bid accuracy, increase asset utilization and reduce the financial consequences of forecast error.

AI becomes useful when it is connected to the data, rules and decisions that define utility operations.

Theme IV

The “all of the above” energy mix must operate as a portfolio

A renewable energy operator with wind farms across multiple geographies faced this problem precisely. Its turbines came from several manufacturers, each with a separate proprietary monitoring system. A centralized monitoring cockpit was developed to consolidate data from those systems into a unified, real-time view of asset performance.

The platform did more than aggregate information. It incorporated service-management capabilities that automatically detected performance and connectivity issues, triggered incidents and supported faster operational response.This is the operating model an increasingly diverse energy system requires: a common layer that can observe, compare and coordinate assets that were never designed to work through the same interface.

Digital twins and advanced scenario modelling can further extend this capability.

Utilities can simulate how changes in weather, demand, market prices, outages or asset availability affect the portfolio before making an operational commitment.

In a more volatile energy system, the ability to rapidly compare alternative operating strategies will be more valuable than relying on a single deterministic plan.