Everything About AI
NextFin News - Massachusetts Governor Maura Healey issued an executive order on Tuesday requiring new data centers to win approval from their host communities before they can seek state permits, while binding the facilities to energy, water, and community-benefit standards that effectively force developers to supply or procure their own clean power. The order, announced September 8, 2026, converts a framework the Healey-Driscoll administration published in June into enforceable expectations for any project that wants to build in the state — and it marks one of the clearest signals yet that the AI-driven data-center boom will face state-level gatekeeping rather than an open-door rush.
"Unless a community says yes to a data center, we are saying no," Healey said at the announcement.
The state is home to about 40 data centers, with the largest located in Lowell, and the new rules apply to projects still in the pipeline rather than existing facilities. A Massachusetts Superior Court judge recently ordered the Markley Group to pause its plan to expand its Lowell facility after private citizens sued — a reminder that the political and legal friction around data-center growth predates this order.
The Rule: Local Veto, Clean Power, and a Ratepayer Backstop
The executive order rests on three pillars. First, a developer cannot enter the state permitting process without a community benefits agreement with the host city or town, and non-disclosure agreements — long a staple of quiet data-center negotiations with municipalities and utilities — are barred. Second, projects must adhere to the energy and water standards set out in the June framework, under which developers are required to supply or procure 100 percent clean energy under Massachusetts' Clean Energy Standard so their load does not push up electricity costs for nearby residents. Third, the order creates a Ratepayer Protection Fund that returns money to customers if a data center's electricity costs are not fully covered by the company.
The June framework, titled the "Statement of Expectations for Responsible Data Center Development and Operations in Massachusetts," states plainly: "New data centers should not exacerbate energy supply availability, wholesale energy prices, or greenhouse gas emissions; instead, their development should help accelerate clean energy deployment." The framework also requires developers to pay for grid and water impacts, mitigate air and noise pollution, and hand over operational-footprint and resource-demand data to the state's Energy Facilities Siting Board. It supplements existing siting rules, including Energy Facilities Siting Board review for onsite generation above 25 megawatts and review under the Massachusetts Environmental Policy Act.
The policy shift follows a legislative detour. In November 2024, the Massachusetts Legislature approved a sales and use tax exemption — codified under M.G.L. c. 64H, § 6(zz) — to encourage data-center development. In June, Healey pressed pause on accepting applications for that incentive, saying stronger protections were needed before communities could be exposed to the costs of hyperscale facilities. Tuesday's order is the administration's answer to what those protections should look like, and it also directs executive agencies on how to evaluate proposals while the administration pushes for additional legislative authority in some areas.
Why Massachusetts Is Doing This: The Grid Math Behind the Politics
The politics are straightforward — Healey, a Democrat, is running for reelection, and her Republican opponent, Mike Minogue, has argued the state lacks the power capacity to absorb more data centers. "At this time, we can't handle more energy demand," Minogue said outside the State House on Tuesday. But the policy is rooted in grid economics that extend well beyond one campaign.
Data centers are load monsters. The International Energy Agency estimates they consumed 1.5 percent of global electricity in 2024, and a typical AI-focused data center draws as much power as 100,000 households, with the largest under construction consuming 20 times that. The U.S. Department of Energy projects data centers could account for up to 12 percent of American electricity demand by 2028. Pew Research counted more than 3,000 operational data centers in the United States as of April 2026, with more than 1,500 in development worldwide — a pipeline large enough that even a fraction landing in one region would reshape its power market.
In New England, the grid operator ISO New England reported in its 2026-2035 forecast, published in May, that net energy use is set to rise after decades of decline, driven by the electrification of heating and transport. Peak demand is climbing with it: the summer 2026 peak is forecast at 25,228 megawatts, and the winter peak is projected to reach 26,849 MW by 2035. That is before any large wave of data-center load arrives.
The capacity market tells part of the story. ISO-NE's Forward Capacity Auction for the 2027/2028 commitment period cleared at $3.580 per kilowatt-month in 2024, up from $2.590 in the 2023 auction for 2026/2027 — a roughly 38 percent increase that shows the region's supply cushion is tightening. The trajectory has been uneven but upward: the 2022 auction for 2025/2026 cleared at $2.591 per kW-month in the rest-of-pool zone, while the 2021 auction for 2024/2025 hit $3.980 in the Southeast Massachusetts zone, a level that had policymakers worried about localized scarcity even before the AI buildout entered the conversation. Wholesale energy in New England averaged $65.88 per megawatt-hour in 2025, with $11.3 billion traded across the region's markets. A new data center that draws power from the common pool without adding supply does two things at once: it bids up the price of capacity and energy, and it forces other customers to absorb the difference. Massachusetts' "bring your own clean energy" model is designed to sever that link — the facility's load must be matched by its own clean generation or procurement, so the marginal cost of AI compute does not land on household bills.
For now, the near-term load impact is modest. ISO-NE's large-load forecast, published in May 2026, identified just two proposed large-load projects in the formal study phase across New England, with combined demand of up to 285 megawatts, and said they would not affect overall system demand before the winter of 2027/2028. That gives the state room to write the rules before the wave hits — which is precisely what the order does.
The state is also building supply in parallel. On March 16, 2026, Healey signed an executive order directing Massachusetts to support the development of 10 gigawatts of new energy resources by 2035, including virtual power plants, and set a goal of 5 GW of energy storage online or under development in the state by the end of 2035. The data-center order and the supply-order are two halves of the same policy: cap the unmanaged demand, and force the new load to pull new clean supply behind it.
The Second-Order Effect: A Template Other States Will Copy
The first-order reading is simple: Massachusetts is protecting its ratepayers and its climate goals from a power-hungry industry. The second-order effect is that the state is writing a template other jurisdictions will follow, and that template changes who wins and who pays in the AI power buildout.
This is a structural shift, not a cyclical political reaction. A cyclical move would be a temporary pause or a one-off fee that reverts once the political pressure fades. What Massachusetts has done is change the permitting architecture itself: it moved the community veto ahead of the state permit, attached clean-power procurement as a condition of entry, and created a standing fund to claw costs back from developers. Those are durable changes to the rules of the game, not a short-term adjustment. The evidence is in the mechanism — the order binds the executive branch's evaluation process and directs agencies to evaluate existing regulations for alignment, which is how a policy becomes a regime rather than a headline. Other states are already moving in the same direction: North Carolina's Durham County voted in August 2026 to pause data-center development, and legislatures across the country are weighing cost-accountability and reporting rules for large loads.
By requiring developers to "bring your own clean energy," Massachusetts is effectively mandating a form of behind-the-meter or directly contracted generation tied to the load. That shifts the economics of data-center development away from utilities and merchant generators selling into the regional pool and toward project developers that can finance and contract clean power at scale. It advantages large technology companies and well-capitalized developers that can sign long-term power purchase agreements with wind, solar, and nuclear operators — and it disadvantages smaller players that rely on buying power from the grid.
The winners in the power complex are the generators that can deliver always-on, carbon-free electricity under long-term contracts. Nuclear operators and independent power producers with clean fleets have already been the primary beneficiaries of the AI power theme: Constellation Energy's chief executive has told investors that projected hyperscaler spending for 2026 is nearly 75 percent higher than the prior year and "continue to be revised upward," while NextEra Energy reported second-quarter revenue of $7.5 billion, up from $6.7 billion a year earlier, with its resources division supplying data centers. But a state-level "bring your own" mandate adds a wrinkle: the value accrues less to whoever owns the plant and more to whoever can bundle generation with the load at the right site. Geography becomes a moat.
The interconnection queue shows why that matters. As of January 2026, about 14,000 megawatts of grid-scale wind, solar, and battery projects were proposed in the ISO-NE interconnection queue — nearly 100 percent of the proposed resources in the region. Almost all of it is clean. The constraint is not the lack of clean supply; it is the speed of interconnection and transmission, and the willingness of host communities to accept both the data centers and the new lines that serve them. Massachusetts' local-approval requirement puts the community veto before the state permit, which means the queue can clear only as fast as local politics allow.
There is a market signal embedded in the order, too. Power stocks that rode the AI-electrification theme — nuclear operators, independent generators, and grid-equipment makers — have traded on the assumption that data-center load growth is a rising tide that lifts every generator in a region. A patchwork of state-level "bring your own" mandates fragments that assumption. The companies that win are the ones that can originate, finance, and interconnect clean generation at specific sites, not the ones that simply own capacity in a regional pool. The trade becomes more project-finance than commodity, and the cost of capital for projects sited in states with uncertain rules rises relative to those with clear, bankable frameworks.
The Counter-Thesis: Gatekeeping Could Cost More Than It Saves
The strongest argument against the order is that it trades one cost for a larger one. Requiring each data center to finance its own clean generation and grid upgrades raises the capital cost of every project, and those costs do not vanish — they are baked into the price of the compute the facility sells. If Massachusetts prices data-center load out of the state, the load does not disappear; it moves to jurisdictions with fewer strings, such as Virginia, which already hosts 665 data centers, or to states actively courting the industry with gas pipelines and nuclear development. The state then loses the property tax base, the construction jobs, and the innovation-economy spillovers, while its climate goals are no better served — the emissions simply get exported.
There is also a reliability argument. A data center that brings its own generation and can operate as a grid-supportive resource — shedding load, discharging batteries at peak, and coordinating with ISO-NE — is an asset to the system. But a facility that is walled off from the grid by strict self-supply rules may be less available to provide those services when the system is stressed. The order's requirement that centers coordinate with the grid operator and act as grid-supportive resources is meant to address this, but the tension remains: the more a project is designed to be self-sufficient, the less it can be called upon to help neighbors.
The falsifying signal for the pro-regulation view is concrete: if data-center investment in Massachusetts falls to near zero over the next 18 to 24 months while neighboring states in New England and the Mid-Atlantic absorb the projects without comparable ratepayer protections, the order will have achieved distributional fairness at the cost of economic activity — and the template will prove less copyable than it looks. Conversely, if projects proceed with signed community benefits agreements and clean-power contracts, and wholesale capacity prices in the region stabilize rather than accelerate, the Massachusetts model will look like the blueprint.
What Comes Next: Base Case, Upside, and Downside
Base case: The order slows but does not stop data-center development in Massachusetts. Projects already in advanced planning renegotiate with host communities, attach clean-power procurement, and move forward on a longer timeline. The state's roughly 40 existing data centers continue operating under prior rules, while the pipeline thins and re-forms around larger, better-capitalized developers. Wholesale capacity prices in New England continue to firm but do not spike, because the largest loads remain years away.
Upside: Massachusetts becomes the model for states managing the AI buildout. The "bring your own clean energy" requirement accelerates interconnection of the 14,000 MW of proposed wind, solar, and storage in the ISO-NE queue, and the Ratepayer Protection Fund proves unnecessary because projects are fully cost-covered. Data-center developers compete to offer the most attractive community benefits packages, turning local approval into a marketable asset rather than a hurdle.
Downside: The local-veto and self-supply requirements price marginal projects out entirely. Investment migrates to states with fewer conditions, Massachusetts loses the tax base and jobs it sought, and the clean-energy queue stalls because the offtake it was counting on never materializes. The Ratepayer Protection Fund is triggered, and the state is left holding costs it intended to avoid.
The signals to watch are the pipeline itself — the number of data-center projects that submit community benefits agreements and clear Energy Facilities Siting Board review over the next four quarters — and whether ISO-NE's large-load forecast is revised upward as those agreements are signed. A second signal is the capacity auction: if the next Forward Capacity Auction clears materially above $3.580 per kilowatt-month, the supply constraint the order is meant to manage is tighter than the state assumed. A third is legislative: whether the Legislature codifies the framework into statute or leaves it as an executive-branch framework that a future administration could revise.
The Massachusetts order is not really about data centers. It is about who pays for the electricity that artificial intelligence consumes — and the state has decided the answer is the developer, the community, and the customer, in that order. Whether that formula keeps the industry in Massachusetts or simply moves it down the interstate is the bet the rest of the country is now watching.
Explore more exclusive insights at nextfin.ai.
