Did the University of Memphis just settle the xAI air fight?

Here we go again!

Have you seen the headlines from every major news outlet in the area that say “XAI has minimal impact on Southwest Memphis air quality?”

All the pro-AI accounts on X had a field day retweeting.

Well, let’s tear that story apart, shall we?

On Tuesday, a team from the University of Memphis told the City Council that xAI didn’t make the air worse. Here’s why that study don’t mean shit.

Back when I was running for Congress, I wrote a long piece about the air in South Memphis. I wasn’t trying to decide if xAI is good or bad. I wanted to know one thing – what has anyone actually measured?

The answer was almost nothing. That’s why both sides could keep waving studies at each other.

So at this meeting Dr. Chunrong Jia and his team from the U of M School of Public Health brought new findings to the council. I read the whole thing. Here’s what’s in it.

After xAI fired up its gas turbines on September 1st, 2024, the team found no real drop in air quality nearby. They also found no real rise in ground temperature. Their results slide says pollution levels mostly went down. Three pollutants dropped enough to count as a real change – ozone, sulfur dioxide, and carbon monoxide.

Now look at how they got it.

There’s a line on their slide, in capital letters, that nobody has quoted. “There are NO field monitoring data for air quality in Southwest Memphis.”

That’s why they used satellites. Not because satellites are better. Because there was no regular monitor on the ground down there during the year they studied. That hole is the exact thing I spent four thousand words on in July.

Their limitations slide keeps going – satellite data isn’t a real measurement, it’s a stand-in. The image squares are too big to tell one smokestack from another. And they couldn’t study anybody’s health at all – their slide blames a lack of health data for those neighborhoods.

I have record requests out to get these numbers – nothing has come back yet.

Their last slide asks for three things. Air monitors on the ground, a health study and a look at what all this does to people’s money and jobs.

That’s pretty much my list from my original essay. Fancy that!

So just from that – any moron should be able to see that these headlines are misleading. If it was definitive – why do they need more info??

So, let’s now talk about the problems with the study in general!

The first problem is the date. The study covers September 1st, 2024 through August 31st, 2025. That’s it. Anything after that is outside the study. So, what’s been happening for the last year? Has anything changed? Were they even running the data center at full capacity when the study was done? We don’t know.

The second is the address. This study is for Colossus 1, on Paul R. Lowry Road, and the neighborhoods are Boxtown and Riverview. It doesn’t touch Southaven, where the turbines the NAACP sued over sit at Stanton Road in Mississippi – a different state with a different agency in charge – and what that power plant produces doesn’t respect state lines. The first of those turbines showed up on August 1st, 2025 – three weeks before this study’s clock ran out.

The third is simple math. The study looked at averages. An average smooths out the bad hours.

When TIME paid University of Tennessee researchers to check the same gas by satellite, the average barely moved, about 3%. But the worst hours inside Boxtown ran 9% higher. In the squares right around the plant, the worst spikes ran 79% higher. Both studies are correct. They’re answering different questions.

The fourth reason is the big one. There’s a second study, and nobody’s putting it next to this one.

Back in February, a group called EmPower Analytics did a health study for the Southern Environmental Law Center. They looked at the 41 permanent turbines proposed for Southaven. They ran the EPA’s own health cost model on it.

Their number is 30 to 44 million dollars a year in health damage. Over the plant’s 30-year life, 588 to 862 million.

And this is what nobody’s talking about – they found Shelby County carries 63.8% of the total exposure to that Mississippi plant’s pollution! The worst neighborhood increases are in Whitehaven and Westwood.

So one study looked at Memphis, looked backward one year, and found no damage. The other looked at Mississippi, looked forward thirty years, and found the damage shows up in Memphis.

This is a classic case of cherry-picking data. No one is looking at this situation as a whole. And reporters are not pointing this out – and it’s so easy to see!

Everything I asked for in July is still missing – except one thing.

The monitor I said wasn’t running is running now. The county’s ozone readings started showing up on the federal air site on August 16th with no announcement from anybody, and I wrote that up on September 9th. The county says the carbon monoxide part of it still isn’t working. I followed up with SCHD literally yesterday about some questions I have about this and got a reply asking me if I was with the press.

And there’s still no count of how many turbines are running at Colossus 1. The Southern Environmental Law Center counted 35 from the air in March 2025, and the county used that number in June. Fifteen got permitted on July 2nd, 2025. Satellite pictures from the day before showed 24. On July 17th, 2025, the company said all the temporary ones had already been “decommissioned.”

Then on September 1st this year, MLGW’s president told the council he doesn’t know how many are running. He said when the second 150 megawatts of grid power shows up this fall, “the rest of those temporary generators will move off.”

The rest. Fourteen months after the company said they’d all been decommissioned.

And this goes back to what I wrote last night about the 11 turbines that supposedly were turned off down at the Southaven plant that powers Colossus II. No one really knows if any of them are “decommissioned” or “moved off” – and it matters because those aren’t the same word, and neither one is spelled out in any record I can point you to. But nobody’s put out a count of the temporary ones since July 2025, and the utility that powers the place says it doesn’t know.

One more thing and I’ll stop.

The permit for those fifteen turbines is called 01156-01PC. It’s the only permitted turbine fleet in Memphis. It expires on January 2nd, 2027.

That’s fifteen weeks out.

The company already asked in March to turn this into an operating permit. The county ran a legal notice about it in The Daily News in April and took comments until early May.

Listen, I don’t know if the people in this entire chain of events, from the city, to the U of M, to the reporters know that they are being misleading – or if they’re just not very curious people. I truly would not be surprised at either of those. I am not saying anyone is acting nefariously – but they could be.

Since I set out to investigate all of this – I have not been after a specific finding. I have just followed the facts. And if I can do it – I just don’t understand why nobody else can.

SMDH

If you like facts-first info on all this, consider leaving a tip on Venmo or PayPal – this stuff takes a lot of time to put together. And check out everything I’ve written with citations and receipts at the xAI Files.

Sources

1. University of Memphis School of Public Health, “Impacts of the xAI Colossus 1 Data Center on the Air Quality and Temperature in Southwest Memphis,” presented to the Memphis City Council, September 15, 2026
2. Commercial Appeal, “SpaceXAI has minimal impact on Southwest Memphis air quality,” September 16, 2026
3. Rachel Hurley, “The air in South Memphis – and the receipts,” July 2026
4. TIME, “Inside Memphis’ Battle Against Elon Musk’s xAI Data Center,” August 2025 – the University of Tennessee satellite work, with the 9 percent and 79 percent worst-hour figures
5. Southern Environmental Law Center, “New study finds proposed xAI gas plant could worsen regional air pollution, cause millions of dollars in annual health damages,” February 16, 2026 – the EmPower Analytics study
6. Rachel Hurley, “How did 68 of xAI’s turbines get into Mississippi with no permit?”, September 15, 2026 – the eleven turbines, and MDEQ’s written answer that it has not inspected the site
7. Rachel Hurley, “An update on the air around xAI’s Colossus 1,” September 9, 2026 – the Riverview monitor
8. Shelby County Health Department air construction permit 01156-01PC, July 2, 2025, annotated – the fifteen permitted turbines and the January 2, 2027 expiry
9. Ars Technica, “xAI data center gets air permit to run 15 turbines, but imaging shows 24 on site,” July 3, 2025
10. Memphis City Council, Doug McGowen’s data center briefing, September 1, 2026, at 5:41:08 – “the rest of those temporary generators will move off”
11. MCAP and CEEJH, “South Memphis hyperlocal air quality monitoring follow-up report,” April 2026

Correction, September 27th, 2026: This essay first said the county counted 35 turbines in June 2025, that photos the day after the permit showed 24, and that the company said the temporary turbines were gone. The 35 was the Southern Environmental Law Center’s March count, which the county repeated. The 24 came from satellite pictures taken July 1st, the day before the permit. The company’s word was “decommissioned.”

Further correction, September 27th, 2026: This essay said the NAACP sued over 69 turbines – its complaint counts 27. It said 63.8% of the people breathing the Southaven plant’s pollution live in Shelby County – the study found Shelby County carries 63.8% of the total exposure. And it said no one was letting people know about the next permit – the county ran a legal notice in April and took comments until May.

Update, September 27th, 2026: A few lines now say who says so, and the line about monitors now says it covers the year the researchers studied.

Everything I have written about AI is in one place at rachelandthecity.com/the-ai-files.