AI’s energy appetite is emerging as a supply chain risk

As artificial intelligence adoption accelerates, supply chain leaders are confronting an unexpected constraint: electricity. According to December 2025 research from Gartner, AI-driven demand is rapidly outpacing global energy infrastructure, creating material risks to supply chain continuity, cost stability, and operational resilience.

Gartner projects that global data center electricity consumption will more than double by 2030, with AI workloads as the primary driver. At the same time, electricity prices are rising at nearly twice the rate of inflation, while the cost of transformers, cables, and critical materials such as copper and aluminum continues to increase. Together, AI adoption, industrial electrification, and population growth are expected to account for roughly 90% of global electricity load growth by the end of the decade, the report noted.

“AI demand is expanding like a high-pressure hose being forced into a century-old energy grid,” the report notes, raising concerns that grid reliability may soon become a strategic constraint rather than a background utility assumption.

From abstract risk to operational impact

For many organizations, energy availability has not historically been treated as a first-order supply chain risk. Kevin Lawrence, a Gartner expert focused on supply chain risk and resilience, tells Supply Chain Management Review that it is time that changes.

“Energy disruptions would equate to similar point-in-time disruptions resulting from weather events,” Lawrence says. “You would quantify the impacts similarly, like downtime of production if there is no failsafe on-site backup power generation in place. Otherwise, it’s less operational and more cost impact, or spike.”

In other words, the consequences of grid instability depend heavily on preparedness. Organizations without backup generation face potential production downtime, while those with safeguards may experience the disruption primarily through higher energy costs.

Detecting grid instability before it hits operations

One challenge for supply chain leaders is that many traditional energy indicators lag actual risk. Lawrence recommends monitoring signals that point to future congestion rather than reacting after prices rise.

“I would start to monitor wholesale electricity prices obviously, but that may be a lagging indicator,” he says. “A leading indicator would be the number of planned data centers in the development pipeline. If you really want to get ahead of this trend, it will require some extra effort to focus resources on tracking what is happening in the pipeline.”

Electricity prices, he adds, function as a secondary lagging indicator, reflecting demand that has already materialized.

Gartner’s research highlights particular risk concentration in regions with dense data center development, including parts of Virginia, Texas, and California, where grid connection queues can stretch 7 to 10 years.

Elevating energy into the risk landscape

By 2030, Gartner expects energy supply constraints to appear more prominently on the supply chain risk radar, but not uniformly across industries or geographies.

“For some it may [become a top strategic risk], but there are many factors involved for each individual circumstance,” Lawrence says. “I doubt that energy supply is currently on the risk landscape for many supply chains today, so step one is including it in your risk management activities.”

That inclusion starts with defining risk tolerance, establishing meaningful KPIs, and ensuring ongoing visibility into energy exposure across the network.

What delivers the fastest return?

Gartner outlines a three-horizon mitigation strategy—immediate, intermediate, and long-term—ranging from financial hedging to structural network redesign. Among near-term actions, Lawrence says Power Purchase Agreements (PPAs) often deliver the fastest financial benefit.

“In short, PPAs can deliver quick ROI, but they are simply a stopgap to hedge against the financial impact of rising electricity prices,” he says. “They don’t do much to protect the operation from physical disruptions.”

Tools such as Energy Management & Optimization Systems (EMOS), on-site generation, and storage investments begin to reduce exposure more fundamentally, but only when evaluated as part of a coordinated, long-term strategy rather than isolated initiatives.

“These various actions should be evaluated strategically and collectively as part of a long-term plan versus independent actions,” Lawrence says.

Modeling volatility with better data

AI itself can play a role in managing the risk it helps create, particularly through scenario modeling and forecasting. According to Lawrence, accurate modeling depends on integrating multiple layers of data.

“Energy prices, energy supply and demand, market and regulatory data, weather, and then more supply chain-specific data such as facility energy usage and production schedules, economic data, and even trend data on related technologies,” he says. “The more granular of these the better to prescribe a scenario of inputs and outputs.”

This convergence of energy and supply chain data is becoming essential as volatility increases across both domains.

Rethinking site selection and long-term planning

Energy availability has traditionally been treated as a narrow feasibility question in site selection: can the grid support projected demand? Lawrence argues that approach is no longer sufficient.

“The aperture needs to widen and the horizon extended beyond such a limited question,” he says. “Rising electricity demand is increasing not solely because of data center growth [but because of] electric vehicles, IoT proliferation, climate change implications driving heating and cooling demand, to name a few.”

Gartner’s research also warns that by 2035, regulators may impose taxes on private electricity networks (PENs), complicating long-term investment decisions. Lawrence advises CSCOs to factor this possibility into planning now.

“If there is a tax on the horizon, you can start accounting for it now to build a financial runway to absorb it or offset it,” he says. “If a PEN is determined to be the best course of action today to mitigate risk, a tax in 10 years may be more of a road bump than a major concern.”

Policy, incentives, and the widening gap

Finally, national policy shifts, particularly incentives for domestic AI infrastructure, could further strain energy systems if not paired with grid investment.

“Ideally, the incentives for domestic AI infrastructure are paired with incentives to invest in grid improvements,” Lawrence says. “Without such considerations, the gap will widen between supply and demand.”

A growing constraint CSCOs can no longer ignore

Gartner’s report concludes that AI-driven electricity demand is no longer a distant infrastructure issue. Without alternative energy strategies, stronger resilience planning, and better integration of energy risk into supply chain decision-making, rising costs and intermittent power access will increasingly threaten operational continuity and margins.

For CSCOs, the question is not whether energy risk will intersect with supply chain performance, but how prepared their organizations are when it does.

SC
MR