How AI Data Center Expansion Shifts Infrastructure Costs

The rapid growth of AI data centers is shifting wealth from residential utility customers to tech companies and utility shareholders. While the public narrative centers on innovation, the underlying mechanics are different: utilities are incentivized to overbuild infrastructure based on speculative demand, and they pass the financial risk to the public. This situation shows how the move fast and break things culture of tech clashes with the regulated power grid. If the AI market falters, the average consumer is left with permanent cost increases. Understanding these feedback loops is necessary for anyone tracking energy policy, corporate accountability, and the long-term economic effects of the AI industry.

The Build-to-Profit Feedback Loop

The core issue in the power sector is a misalignment of incentives between utilities and their customers. Utilities are profit-seeking businesses that generate returns by building large-scale infrastructure. When data center demand forecasts rise, utilities do not just react; they increase their capital spending.

As John Farrell notes, utilities make money by spending money. By using speculative demand to justify massive transmission and generation projects, they secure a return on investment that exceeds the risk-adjusted market rate. Because these companies operate as monopolies, they face little pushback from regulators who often approve these projects without deep scrutiny. The result is a system where the utility profits regardless of whether the projected data center demand actually happens.

There is definitely a huge incentive for utilities to make that kind of investment if they can get away with it because that is how utilities make money they make money by spending money by building big infrastructure.

-- John Farrell

The Asymmetry of Risk

A major, overlooked consequence of the data center boom is the permanence of rate hikes. Analysis by Cathy Kunkel shows that grid operators use forward auctions to plan capacity years ahead. When demand is overestimated, the system locks in costs for infrastructure that may never be fully used.

If the AI bubble bursts and these data centers are not built, the utility does not take the loss. Instead, they are left with undepreciated costs, which is the debt from building power plants that are no longer needed, and these costs are passed to the remaining ratepayers. This creates a heads I win, tails you lose dynamic for the utility, where the consumer pays for corporate speculation.

The Illusion of Off-Grid Independence

Tech companies often argue they can bypass the grid by building their own power plants. However, this is not an isolated solution. Building private natural gas plants creates a ripple effect: it tightens the supply chain for gas turbines, which raises costs for everyone else, and it creates reliability risks if those private plants fail and force the data center to draw power from the public grid.

If you are building an off grid natural gas plant you are still building a natural gas plant which means you are like contributing to the fitting up of the cost of natural gas plants around the country.

-- Cathy Kunkel

The Power of Novel Precedent

The Haymarket transmission line dispute shows how to disrupt established corporate-utility patterns. By forcing a utility to bury a transmission line, the community proved that the line extension policy, which requires the party requesting power to pay for the infrastructure, could be applied to Big Tech. While the community had to compromise, they established a record that the data center industry is a unique customer class that triggers specific, expensive infrastructure needs. This precedent is the basis for future resistance: it shifts the argument from not in my backyard to pay for your own extension cord.

Key Action Items

• Map Local Utility Filings: Over the next quarter, look at your state Public Service Commission filings for Certificates of Need. Identify if your utility is justifying new infrastructure based on speculative data center demand.
• Audit Your Bill Structure: Use the next 3 to 6 months to document your bill specific charges. If your utility prevents year to year comparisons, treat this as a signal to engage with local advocacy groups that track these transparency issues.
• Apply the Line Extension Argument: In local zoning or regulatory hearings, push for the application of line extension policies to data center developers. Demand that the cost of new infrastructure be paid by the developer, not the ratepayer. This is a long-term investment of 12 to 18 months that creates a durable policy moat.
• Support Local Coalitions: Focus on community organizing rather than just technical testimony. As Kunkel notes, the most effective tool is demonstrating widespread political risk to local officials. This pays off immediately in terms of visibility and long-term in terms of electoral accountability.
• Challenge Off-Grid Narratives: When tech companies claim they are building their own power, push back on the impact of their supply chain usage, such as turbine shortages, and the lack of grid reliability protections. This requires sustained pressure on regulators to force these entities to account for their systemic footprint.