Massachusetts just became the latest state to tell data center developers: bring your own power, and make it clean. Gov. Maura Healey signed an executive order requiring any data center larger than 25 megawatts to supply clean energy or pay into a ratepayer protection fund, according to TechCrunch AI. As TechCrunch AI reports, it’s the third state in three months to crack down on an industry that politicians used to compete to attract.
This is significant because it marks a sharp reversal. A few years ago, states showered tech companies with tax breaks and incentives to land these facilities. Now they’re writing rules to slow them down, and voters are the reason.
What the order actually requires
Healey’s order puts real conditions on large builds. Here’s what developers now face in Massachusetts:
- The 25MW trigger. Any data center above 25 megawatts of peak demand must bring its own power and guarantee it meets the state’s clean energy requirements.
- 100% clean, not a slice. The state’s standard normally asks industry to source only a portion of power from approved sources like wind, solar, and hydro (at least 40% by 2030, rising over time). But the governor’s office clarified to TechCrunch AI that data centers must meet 100% of their electricity demand with clean generation.
- On-site preferred. Healey wants developers generating that power on-site. If they can’t, they’ll need to fund new generation nearby or pay into the ratepayer protection fund.
- No secret deals. The order directs communities to avoid signing non-disclosure agreements with developers.
- Tax break on pause. The governor is pausing applications for a data center sales tax exemption that took effect just last month, giving regulators time to implement the rules.
What stands out here is the 100% clean requirement. A typical commercial user only has to hit the rising state percentage. Data centers now carry a heavier load than almost any other industry in the state.
Why this matters for AI
Data centers are the physical backbone of the AI boom. Training and running large models eats enormous amounts of electricity, and developers have been racing to build capacity. These rules raise the cost and complexity of that buildout, at least in states taking this route.
Massachusetts isn’t alone. TechCrunch AI notes the pattern across recent months:
- July, New York halted construction of new data centers 50 megawatts or larger.
- August, Texas required all new data centers to submit to audits by the public utility commission and grid operator ERCOT.
- September, Massachusetts set the 25MW clean power mandate.
Three different states, three different mechanisms, one direction of travel. The status quo of “build anywhere, we’ll pay you to come” is gone in a growing number of places.
The driver is public sentiment. Residents worry about strain on the grid and higher utility bills when a single facility can draw as much power as a small city. Politicians are responding to that pressure directly.
The industry is pushing back
The tech side isn’t sitting still. TechCrunch AI points to Leading the Future, a pro-AI super PAC funded by Marc Andreessen, Ben Horowitz, and Greg Brockman, that’s buying ads to sway voters in battleground states ahead of the midterms. That tells you how seriously the industry takes this shift. When billionaires start funding political ad campaigns, the fight has moved from permitting offices to the ballot box.
What to watch next
If you’re building, financing, or siting AI infrastructure, the location math just changed. A few things to track:
- Whether other states copy the Massachusetts 100% clean model or the New York and Texas approaches.
- How developers respond: on-site generation, nearby projects, or fund payments.
- Whether the super PAC spending actually moves voters, or hardens the backlash.
The era of states bidding against each other to host data centers is fading. In its place, a patchwork of power mandates, audits, and moratoriums is forming, and it’ll shape where the next wave of AI compute gets built. For the full details, see the original report from TechCrunch AI.