Three Hours, One Megawatt: What Was Said at the September 9 Town Hall

The University of Michigan and Los Alamos National Laboratory held a town hall on September 9, 2026 at the Marriott at Eagle Crest. It ran nearly three hours. The full recording is public, and every timestamp below links directly to that moment in the video so you can hear it for yourself.

Two notes before we start. Quotations come from the recording’s automatically generated captions, which contain transcription errors; we have corrected obvious mis-hearings and linked every quotation to the moment it was spoken, so you can check us. And the author of this site attended and asked questions that evening. Where that matters below, it is written in the first person and you should read it as a participant’s account rather than a neutral one.

Who was there, and who was not

The panel had impressive resumes. It did not have the authority to change, alter, or stop the project. Residents asked about this directly, and the panelists confirmed it: the people who can make that kind of decision are the University Regents.

The only regent present was Paul Brown, who has publicly called on the University to pause the project.

Paul Corliss	Assistant Vice President for Public Affairs and Communication; also facilitated the event
Steven Ceccio	Professor of engineering, College of Engineering; project leader
Rob Green	Project director
Patricia Hurn	Interim vice president for research and innovation
Chris Kolb	Vice president for government relations
Drew Smith	Interim vice president for finance
Khalil Rahal	Director of economic development, DTE Energy

Absent: the current University president, the incoming president, every regent except Paul Brown, and anyone from Los Alamos National Laboratory.

That last absence is worth sitting with. By the University’s own figures, roughly 90% of this facility belongs to Los Alamos. At a public meeting held to answer this community’s questions about it, Los Alamos sent nobody.

Several residents told the panel they came across as dismissive and unprepared. We think that criticism was fair — not as a comment on anyone’s character, but because the answers contained so little information. What follows is the substance.

The 90/10 split, confirmed on the record

This was the most important thing said all evening, and it nearly did not get said at all.

I asked the panel about the division of capacity between Los Alamos and the University — roughly 90% and 10%. The question was met with blank looks from most of the panel. Only after moderator Paul Corliss turned to project leader Steven Ceccio for confirmation did we get it: roughly a 90/10 split, with the large majority of capacity allocated to Los Alamos.

Separately, at 2:14:11 another resident asked what percentage of the facility would be Los Alamos and whether it is classified space, saying plainly that she was there for the data. She did not get a percentage. The answer restated that the facility is for computational work, the exchange became heated, the moderator asked the room for order, and the meeting moved on.

None of this was secret. The University told the State of Michigan in December 2024: 100 megawatts for Los Alamos and a dedicated 10-megawatt academic center for the University’s non-classified research. Its own FAQ describes a 240,000 square foot federal building including classified research alongside a 50,000 square foot academic building. The answer existed, in writing, for twenty-one months. It took a direct question and a follow-up from the moderator to get it spoken out loud in Ypsilanti Township.

The two words that were not said

Across roughly 27,000 words of transcript, the word “megawatt” appears once. It was not spoken by anyone on the panel. It was spoken by me, at 1:36:26, making the point that measured against the average American data center, a 100-megawatt facility sits at the very top of the range.

The word “gallon” does not appear at all. Not once, in three hours, at a public meeting about a facility whose water consumption has been this community’s most persistent question for over a year.

That is the finding. A postcard to every household, a signed letter from the Director of Los Alamos, and a three-hour public meeting have now produced no operating figure for power and no operating figure for water.

“Still in the planning phase”

Question after question came back to the same answer: the project is still in the planning phase and the details have not been worked out.

Data centers are being built at record speed across this country. The technologies involved — cooling, backup generation, heat rejection — are well understood and thoroughly documented. A $1.25 billion project with a completed design phase, a purchased site, and a substation timeline does not lack specifications. Two things said publicly show as much.

The cooling technology is already named. The project’s own materials state it will use direct-to-chip computer cooling, which they describe as 10 to 15% more efficient than the air-cooled systems used in typical data centers. That is a specific engineering choice with a specific efficiency figure attached. Either the cooling design has had considerably more thought than the panel offered, or that claim is not backed by analysis. Both cannot be true.

The backup generation is already sized. The project has committed publicly to backup capacity at 10% of load, running on hydrotreated vegetable oil. To make that commitment you must have calculated 10% of a load you know, then specified generator sets that deliver it and are certified for that fuel. That is not planning-phase vagueness; it is a design decision. Unless the number was chosen to reassure residents rather than to describe equipment.

On cooling and water

This was the most technically candid stretch of the evening, and the panel deserves credit for it.

At 47:00 the panel explained that cooling can be done several ways: a closed-loop system that uses no water, or an open system using reclaimed wastewater from the treatment plant rather than potable water. The reclaimed-water option, they said, had been explored with the utilities before the moratorium went into place.

At 52:04 came the sentence that matters most: closed-loop cooling essentially uses more electricity but no water. The panel described it as one option and called this still a big decision about how we might do this.

In describing the reclaimed-water option, project director Rob Green said the wastewater would otherwise be dumped back into the Rouge River farther downstream. The river beside this property is the Huron, not the Rouge. It may well have been a slip of the tongue. For a room full of people who live on that watershed, it did not land that way.

So the cooling design is not settled — which matches the FAQ’s more careful phrasing, that the University is reviewing designs to consider closed-loop cooling. It also states the trade-off plainly: no water means more electricity. Residents are entitled to know which way that decision goes, and what each option costs in gallons and in megawatts, before it is made rather than after.

On electricity and rates

Early in the presentation the panel addressed concerns about electricity. The power draw was described as relatively small compared to these other large data centers, and because the project will build its own substation, it should not have any impact on both your service or your rates. (20:51)

Two observations. “Relatively small compared to these other large data centers” is the same unspecified comparison that appears in the project FAQ — small relative to what was never stated. And building your own substation addresses one route by which costs reach ratepayers. It does not address generation capacity, transmission, or how a large new firm load affects system costs generally.

Late in the evening a resident raised exactly that distinction. The exchange beginning at 2:29:05 is worth hearing in full. After a resident said rates would rise because of the substation, the response was that the project will pay for its own substation and that DTE will not sign a contract with anybody that will raise your rates because of that contract. A resident immediately replied that this is not the same thing — and they were right. A commitment that one contract will not raise rates is narrower than a commitment that rates will not rise.

DTE then described the site as served by a dedicated industrial substation the project pays for, separate from the distribution system serving homes, and cited transmission-level service. (2:29:56) That is a meaningful answer on reliability. It is not an answer on cost allocation, which is decided at the Michigan Public Service Commission.

On jobs

A resident asked directly how many permanent jobs would go to residents of Ypsilanti Township. The answer began:

you know legally I’m not sure we can say that jobs will go to Ypsilanti township residents

Panel response, 1:55:59

We think that is an honest answer, and we are not criticizing the person who gave it. Hiring commitments to a specific municipality do raise real legal questions.

But it is the answer, and it should be read alongside how this project has been described publicly. The panel then restated the familiar figures: 30 to 50 technical support jobs at the facility, 200 research jobs requiring advanced degrees, and roughly 300 construction jobs. (1:56:06)

Note what that ordering makes clear. The 30 to 50 are described as at the facility. The 200 research positions are not. That distinction has been the central ambiguity in this project’s employment claims since December 2024, and the town hall did not resolve it.

On the site

The panel said that of the 144 acres, roughly three-quarters will not be used for construction. (18:04)

That is a real commitment if it is written down. Our question is unchanged: by what instrument? A conservation easement and a recorded cap on developed capacity would make it binding. A slide does not.

What we take from the evening

Some genuine information came out of this meeting. The 90/10 split was confirmed out loud. The cooling trade-off was explained candidly. The jobs answer was more honest than the marketing. DTE addressed reliability in specific terms.

And after three hours, this community still cannot answer the two questions it has been asking longest: how many megawatts, and how many gallons a day.

Those numbers exist. They are in a design basis document and a load study. They are not classified. Publishing them would take an afternoon and would do more for this community’s confidence than another public meeting.

There is a second town hall on October 1. The same questions will be waiting.

---

Quotations are from the automatically generated captions of the public recording of the September 9, 2026 town hall; each timestamp links to that moment so it can be verified. Capacity, budget and square footage figures are from the University’s December 2024 application to the Michigan Strategic Fund and its LANL FAQ; see our sources page. If any quotation or attribution here is inaccurate, please tell us and we will correct it and note the correction.