When AI Strains the Grid, Who Pays the Bill?

After the affordability crisis, AI may be the most popular topic in public life right now. It’s mentioned in campaign speeches, policy debates, headlines across industries and family dinner conversations. Whatever your personal view of AI is (enthusiasm, skepticism, doomsday or something in between), its rapid growth is no longer theoretical. It is already reshaping how states think about infrastructure, energy demand, and household costs.

One of the less discussed consequences of AI’s expansion and integration to daily life is the strain it places on the electric grid and water supplies. The data centers that power AI systems are extremely energy-intensive, and that demand has real implications. 

In recent months, multiple AI-related data center proposals have surfaced across the state. In Bloomfield, a million-square-foot data center has been proposed for former tobacco farmland, promising investment while raising questions about land use and energy demand. In Hartford, plans are moving forward for a $90–$100 million Connecticut Center for Applied AI, positioning the city as an emerging regional technology hub tied to workforce development and industry collaboration.

These projects may bring economic opportunity to our state. But they also raise a fundamental question: who pays when demand on the grid increases? Without clear policy guardrails, the answer risks defaulting to households already feeling squeezed by rising utility bills and broader cost-of-living pressures, as we discussed in our last post. 

Last week, just next door in New York, Governor Kathy Hochul addressed this reality directly in her State of the State. Her newly announced Ratepayer Protection Plan is a comprehensive effort to reduce utility costs and strengthen grid reliability. Notably, one component would require large energy users, including AI data centers, to either pay for the grid upgrades they necessitate or supply their own power when they do not deliver commensurate job or community benefits. 

New Jersey is moving in a similar direction. On her first day in office, Governor Mikie Sherrill signed executive orders aimed squarely at utility affordability, which she campaigned almost exclusively on. One order requires electric utilities to report energy requests from AI data centers, a transparency step intended to give regulators and the public a clearer picture of how rapidly growing AI demand could affect system costs.

Connecticut’s legislative session will open soon, and the policy agendas for the Governor and legislative leaders have yet to be unveiled. As proposals for AI data centers move from headlines to permitting and construction, the choices made by our elected leaders now will determine whether or not rising demand quietly shifts infrastructure costs onto taxpayers and ratepayers later. 

Growth is a policy choice. So is deciding whether families absorb the costs that come with it.