The Download

Illustration: Morning Brew Inc., Photos: Adobe Stock

TL;DR: The data center backlash is getting personal—and no one is feeling the effects quite like John Perkins. The city council member was recalled by voters in small-town Missouri on Tuesday after he backed a more than $6 billion tax break for a data center project. Perkins was years away from reelection—making him one of the first people to be pushed out of office over a data center deal. The vote landed a day after President Donald Trump promised data centers would make towns rich and add lots of jobs—but the actual numbers don’t really back this up.

What happened: Trump has something to say about anti-data center voters: They’ll “end up being backwards and poor,” he posted on Truth Social on Monday. “If they want to be successful and rich, with far lower taxes and jobs all over the place, let Data Reign.”

So, do data centers actually deliver lower taxes and more jobs for these communities? The record so far says mostly no.

Far lower taxes: While a few regions have enjoyed lower property tax rates because of data centers, the tax breaks given to data centers largely negate any benefits.

Forty states have some sort of subsidy or tax break (such as exemptions on sales and utility taxes) for data center projects—and the terms are usually locked in for 10 to 30 years. That’s decades in which a data center contributes far less to the local community while everyone else’s tax bill stays where it was (and in some cases, their electricity bills go up).

For example: In Virginia, for every dollar of sales tax the state let data centers forgo between 2014 and 2023, it got back an average of 48 cents. This was in personal income, corporate, and nonexempt sales taxes, which is all credited to the exemption, according to a 2024 study by the Virginia Joint Legislative Audit and Review Commission. That means it’s losing 52 cents for every dollar—and the state reported a loss of $1.6 billion last year, just in exemptions for data centers.

Jobs all over the place: To their credit, most tax break incentives for data centers require the projects to hire people (those for many other industries don’t). But the actual number of jobs required to qualify is…shockingly low.

The requirements range from just five to 100 (typically permanent) jobs per project. Building a data center can take thousands of workers (over about three years), but running one usually takes just a few dozen—meaning long-term jobs largely never show up. In one instance, a Microsoft data center in Illinois got a tax break of more than $38 million but created just 20 permanent jobs, per CNBC. In New York, a JPMorgan Chase data center promised one permanent job to get nearly $77 million in tax breaks.

Bottom line: It’s a hard sell to ask people with rising power bills to subsidize the richest companies and get little in return. Perkins is an example of what that resentment looks like when communities reach a breaking point and the numbers don’t quite add up. —LC

      

Sponsored By Keeper Security

Your log-ins are currently one sticky note away from a bad time. “Password123” isn’t cutting it, and neither is reusing your dog’s name for every account you own.

Keeper locks down everything—your accounts, your files, your business’s most sensitive data—without making you think about it twice. It generates strong passwords, autofills them everywhere, and flags the ones you’ve been recycling.

For a limited time only, get up to 50% off Keeper personal and family plans and 30% off business plans. Whether it’s just you or your whole team, this is the easiest security upgrade you’ll make all year. Set it up once and forget about it.

Get 50% off the #1-rated password manager before the deal disappears.

Set your AI loose (with rules)

Note: Saira’s off for today, so we have our very first guest column for How to AI thanks to Yash Thakker—an AI practitioner, product builder, and educator with over 12 years of experience building AI-powered products.

I used to babysit my AI assistant like a toddler near a pool. Type a prompt, watch it work, read the output, type the next prompt, repeat forever. Then I found out you can just...not do that. Instead, you can set up a loop: Tell the AI what to do, how to check its own work, and when to stop, then let it run while you do literally anything else. It felt like discovering the washing machine after years of hand-washing clothes in a bucket.

What loop engineering actually is: It’s designing an AI workflow that repeats itself without you clicking “send” every time. The AI cycles on its own: It gathers what it needs, does the task, checks whether it worked, and tries again if not—until it hits a stop condition. You’re not writing better prompts anymore—you’re writing the rules it follows while you’re not watching. For example, the Mozilla team used it to work through hundreds of Firefox bugs in a month (when it normally would have taken them a quarter).

How it works, roughly: Every loop needs four things: a trigger (a schedule, a new email, a specific command), a scope (what it’s allowed to touch), a check (how it knows it succeeded), and a stop condition (when it gives up and pings you). Think less “prompt” and more “job description.”

Try this prompt: “Every time I say ‘run,’ check [a specific inbox folder or shared doc]. Do [a specific action], then tell me what you did and flag anything you’re unsure about instead of guessing.”

That’s a baby loop. Once you trust the loop, look for your tool’s built-in scheduling feature (most major chatbots have one now) so it can trigger itself—at either a certain time or when an action is performed.

What it’s great for: Repetitive, well-defined chores like sorting a messy inbox, drafting the same weekly report, or cleaning a spreadsheet every Monday (anything you’d normally do the exact same way twice).

What it’s bad at: Judgment calls with real stakes. A loop will happily keep “fixing” something in a way that technically satisfies its own check while quietly making a mess, because it has no idea it’s wrong until you tell it. Don’t loop anything you wouldn’t hand to an intern and walk away from for a week.

Start small, watch it once or twice, then let go. —YT

Sponsored By Outreach

Take action. AI no longer just answers questions: It takes action inside your revenue workflows. But organizations still need control, even as AI agents begin to make decisions. That’s where Outreach comes in, providing the control, governance, + security teams need to confidently deploy AI at scale. Book a demo.

The Zeitbyte

A gutsy way to measure your gut health

Morning Brew Inc.

There’s never been a better time to give yourself a literal gut check—whether it’s mail-in microbiome test kits or that viral fart-tracking underwear. But the latest entrant into the business of investigating your digestive tract goes a step further: smart toilets that constantly monitor the makeup of your…well, you know. (And yes, these porcelain thrones come with subscriptions.)

It’s part of a broader boom of tech that measures every aspect of your body, from your sleep score to the quality of your nighttime erections—and it pushes the limits of how much personal data you might be willing to share with a company. These smart toilets, for one, are equipped with cameras pointed into the bowl to enable urine and stool observation, and use AI to analyze what they see. (Though it’s not the only recent tech product that puts a camera where people probably don’t want one to be.)

Plus, there might be a much cheaper way to get a quick picture of your gut health. As one gastroenterologist pointed out to Wired: “If you want to look at your stool, you can just look at it yourself.” And if you don’t want to do that either, there’s always the option to simply know less about your body. —WK

Chaos Brewing Meter: ☕☕/5

• GPT-6 Astra is here. OpenAI calls the new model a “generational leap” and boasts that it can probably book a DMV appointment faster than you.
• Tesla’s Cybercabs have officially hit the streets of Austin—and a federal auto safety regulator already opened a probe into the operation.
• Yet another swarm of OpenAI agents might have gone rogue earlier this year, months before the Hugging Face fracas.
• We’ll likely see a lot more than the foldable iPhone at Apple’s big event next week. Here’s a rundown of what you can expect.
• The second-largest public school district in the US has now banned AI tools on all district-owned devices issued to students.
• Here’s a new way to annoy your coworkers: Talk out loud to your Gmail and Docs with the conversational AI features Google just launched.

^{*A message from our sponsor.}

This time last week...

Readers’ most-clicked story was about this adorable robot duck from Hugging Face.