What AI Demand Reveals About the Utility Model

In the energy industry, it is often said that utilities move when regulation requires it. 

At first glance, this can look like inertia. In reality, it reflects the way utilities were designed to operate. 

The utility model prioritizes: 

• reliability over speed 

• certainty over experimentation 

• regulatory alignment over first-mover advantage 

Every major decision—whether a capital investment, operational change, or new technology deployment—must be defensible within a regulatory framework.  

Cost recovery, compliance, and system reliability are not constraints on the business model; they are the business model. 

Within that context, waiting is often the lowest-risk decision. 

But the environment is changing. 

Electric demand—particularly from AI infrastructure—is growing at a pace that does not align with traditional utility planning cycles. The International Energy Agency projects that data centers could account for nearly half of U.S. electricity demand growth through 2030. [iea.org] 

At the same time, the ability to add new infrastructure is becoming increasingly constrained. 

According to Lawrence Berkeley National Laboratory, U.S. interconnection queues now contain more than 2,000 GW of generation and storage capacity—nearly double the country's existing installed fleet. [emp.lbl.gov]   

The challenge is not simply scale. It is time. 

• Median interconnection timelines have stretched to roughly 4–5 years. 

• Historically, projects often interconnected in less than 2 years. 

• A significant share of queued projects never reach commercial operation. 

This creates a growing mismatch: 

• Demand is accelerating rapidly. 

• Infrastructure expansion remains constrained by permitting, transmission, interconnection processes, and capital deployment timelines. 

The result is a system where load growth is beginning to outpace the speed at which new capacity can be deployed. 

For decades, the industry's primary answer to rising demand was straightforward: build more infrastructure. 

That approach is becoming increasingly difficult to execute at the speed emerging load growth now requires. 

As a result, attention is beginning to shift from planning alone to operations—how existing assets, networks, and resources can be utilized more effectively. 

This raises important questions for utilities and regulators alike: how should operational flexibility be valued, and what role can software-driven optimization play in improving system efficiency before new infrastructure is built? 

This is an important distinction. 

Utilities are not unwilling to innovate. They are operating within incentive structures designed to minimize risk and ensure reliability. The most successful solutions will not be those that attempt to bypass those realities, but those that work within them—improving system performance while supporting regulatory, operational, and reliability objectives. 

AI-driven load growth is not exposing a failure of the utility model. 

It is revealing the limits of a model built for predictable demand growth in a world that is becoming increasingly dynamic.