This year, Lamont seemed to pull back from such open-handed incentives for data centers. A proposal to build a 1.5-million-square-foot hyperscale data center connected directly to the Millstone Nuclear Power Station in Waterford was scrubbed, about which the governor expressed relief.
"Many states unplugged nuclear power and plugged in energy-sucking data centers, driving up prices around the country," he said during his 2026 State of the State address. "Here in Connecticut, we did just the opposite. We extended our Millstone nuclear agreement and said, 'Slow down new data centers,' unless they add more generation as well."
Since 2021, only one company has taken advantage of the tax incentive for data centers. The Cigna Corp. reported investments of over $863 million between 2023 and 2025, and earned $17 million in tax rebates. The project has created five new jobs.
Although there are an estimated 50-plus small to mid-size data centers in Connecticut, there are no so-called hyperscale data centers, which can exceed 1 million square feet. There are also differing views on whether Connecticut needs larger data centers to compete as a center for innovation, or if the disruption, cost and environmental damage is not worth the gain.
As artificial intelligence models are incorporated into all areas of industry, a battle is brewing nationally over how to manage the tremendous increase in infrastructure needed to run those programs. Some states, such as Virginia, have embraced hyperscale data centers, and some legislators have encouraged their growth as both a boon to job creation and a step toward the future.
But the larger the data center, the more power it requires. Other states and lawmakers have urged caution, citing massive power needs and a subsequent rise in the cost of electricity, as well as environmental concerns.
"Data centers are going to be consuming, in fact swallowing, in huge amounts, a lot of energy and electricity simply to operate. And data centers are being built in larger numbers, but also sizes all around the country," said U.S. Sen. Richard Blumenthal, D-Conn., who has proposed legislation called the GRID Act. The act's goal, he said, is "to make sure that consumers are spared the burden of increased costs resulting from the demands placed on electricity and energy production and transmission."
Others are not so sure. Norwich Mayor Swarnjit Singh said, "As a city we are open to data centers."
There are already 15 data centers located in the town, according to datacentermap.com, which tracks their locations. Norwich has its own, dedicated utility company, and Singh said he believes there is enough capacity for more.
"That makes us unique in Norwich," he said. "We have not just land space but also utility infrastructure."
Data centers, he said, "Help taxpayers in the form of the tax base." Singh said he hopes that "Norwich becomes a favorite space for data centers."
The Rise of Data
Daniel H. O'Keefe, commissioner of the state Department of Economic and Community Development and Connecticut's chief innovation officer, said he is excited about AI.
"It's like the emergence of the printing press, where you saw information get more freely distributed in a way that it hadn't before," he said.
"This is now about taking action. The rise of the consumer internet was about information. This is now about taking that information and being able to decision," O'Keefe said. "The concept of knowledge work is being reinvented because now these systems have gotten to a degree of maturity and a degree of confidence that they're now able to actually action, and that, to me, is a step function change."
But O'Keefe said he is under no illusions that as artificial intelligence becomes intertwined with every facet of life, both physical and digital, those increasingly complex models will require more resources.
"The reality is that demand is going to continue because the commercial applications of it are so vast, and, by the way, the economic implications of it are so vast," he said.
The U.S. Department of Energy released in 2024 a report evaluating the increasing power demand of data centers as a result of artificial intelligence. "Data center load growth has tripled over the past decade and is projected to double or triple by 2028," the report says.
O'Keefe's department recommended that the data center tax incentive program "continue due to the positive impact on state revenue." But O'Keefe said he does not believe data centers need to be located in Connecticut for the state to reap the benefit of artificial intelligence. The internet is, after all, in the cloud.
"In commercial production where that latency matters, it's measured in milliseconds, and it only matters in certain specific applications," he said. "If you're a research scientist, the need for a near-zero latency is very low. Because, again, we're talking about milliseconds here, and so blink-of-an-eye kind of latency. Not only that, but you can deliver that kind of lack of latency in a much smaller footprint."
Christine Broadbridge, a physics professor and executive director of the newly unveiled Connecticut State Colleges and Universities Center for Quantum and Nanotechnology, said the state has the infrastructure needed for high-level research.
The state college system has a "robust IT backbone" that "provides high-speed, low-latency fiber-optic connectivity across the state, enabling data-intensive research, simulations, and collaboration essential for both AI and quantum technologies," she said.
The lack of large data centers in Connecticut does not, Broadbridge said, impact the center's work. "High-performance computing resources are readily available locally," she said. And "strong partnerships with Yale University and the University of Connecticut" exist for any computing needs that go above and beyond what is already accessible.
"Its collaborative model allows the lab to remain flexible, quickly plugging into new modalities as they become available," Broadbridge said by email. "When Connecticut adopts a unified statewide approach to high-performance computing, the QNT is well-positioned to integrate seamlessly. Flexibility in using a variety of computing resources enables the QNT to choose the right tool for different aspects of its work."
But not everyone agrees. Fred Carstensen, an economist and professor at the University of Connecticut, has been a vocal advocate for more data centers and IT infrastructure, arguing that the state needs to position itself for the future.
"From what I have seen, Connecticut has a very weak IT infrastructure. We're very weak in the quality of data centers. They're relatively small. They're relatively low quality in terms of their reliability, if there's a power outage or whatever," he said. "You're not going to put the high-end stuff here, because we don't have quality infrastructure."
Data Center Challenges
Ayse Coskun, a professor at Boston University, said modern, AI-focused data centers require far more power than older server stations.
"AI data centers rely heavily on these specialized chips called GPUs, which perform massive numbers of calculations simultaneously to train and run AI models," she said. "Training or building a large AI model such as ChatGPT, Gemini and others that people can query, can require thousands of GPUs operating simultaneously continuously for weeks or even months at a time, so that concentration of computing power translates directly into high electricity demand."
It's not just more chips. The power, she said, is packed into a smaller footprint. Where older server racks might use as much power as a residential home in the U.S., AI data centers can use 10 times that much power.
That electricity is not used just to run the GPUs. That much power use results in heat, and cooling technology requires a water or air supply, which also needs power. As much as 20% of a facility's total electricity usage is for cooling.
All told, the largest hyperscale data centers use 100 megawatts or more.
"To put that in context, that's roughly equivalent to the total electricity demand of a small city," Coskun said. "So, when a hyperscale data center connects to the grid, utilities may need new substations, new infrastructure and transmission lines that deliver the power may require upgrades."
The environmental concerns can also be significant. The water needed to keep large data centers can can reach into the millions of gallons, according to Shaolei Ren, a researcher at the University of California, Riverside.
"In 2030, in the next few years, a new peak, new water capacity request by data centers in the U.S. could be reaching up to 1 billion gallons of water per day," she said. "This amount of water water capacity is, if you put them together, hypothetically, enough to support New York City's water supply on typical days."
Blumenthal's bill would force data centers to provide their own electricity, but Coskun said a data center's power needs are location-specific, to an extent.
"If a data center operates in a region heavily dependent on coal or natural gas, then higher electricity demand can increase fossil fuel generation and emissions as a consequence," Coskun said. "If the data center is in a region with abundant wind, solar, hydro or nuclear power, then the emissions impact might be much lower."
"AI growth and climate policy are now in a way interconnected. How we design these facilities today will influence emission trajectories for possibly four decades," she said.
Political Movements
Blumenthal is not alone in his fight to rein in data center growth. His colleagues U.S. Sens. Alexandria Ocasio-Cortez, D-N.Y., and Bernie Sanders, I-Vt., have proposed a national moratorium on data center development.
But Blumenthal also said he understands the need.
"There is an upside to data centers. First, from the standpoint of our national interest, we need to continue to advance technology and have the kind of communications and knowledge expansion that comes with the data centers," he said. "They're useful to store and transmit information. The data centers are not a problem. In fact, they can be an advantage to the nation and to individual communities."
At the state level, there have been numerous bills proposed to study, encourage or restrain the growth of data centers, though few have been put forward in this year's legislative session. A proposed state bill this year would eliminate the tax incentive for data centers that only Cigna has taken advantage of.
That bill, and the tax incentive it would put an end to, may be moot. As state Senate Majority Leader Bob Duff, D-Norwalk, said, electricity in Connecticut is managed regionally, which makes the issue more complex.
"I don't think there's a tremendous demand here in Connecticut for data centers because our electric prices are high," he said. "I think it's something we should be talking about on a regional basis, not a state basis."
O'Keefe agreed. He said he views hyperscalers as "strategic for the country," but not necessarily appropriate for the region. Large hyperscale data centers create construction jobs, but they are largely autonomous once built.
"Our energy costs are over 100% higher at the wholesale level than the rest of the United States," he said. "Why would you put hyperscaler compute in the Northeast? It just doesn't make any sense to me, and there's other regions that have much more land, easier access to lower cost-energy and easier access to cooling requirements."
© 2026 The Middletown Press, Conn.. Visit www.middletownpress.com. Distributed by Tribune Content Agency, LLC.
