Who Said No?
Issue 455 · Published November 4, 2026
Digitally Literate is a weekly letter about the forces shaping how we read, write, learn, and live in digital spaces. Subscribe to get each issue by email.
Two weeks ago, I asked who gets to say yes. This week, the question flipped. Who gets to say no?
On Tuesday, voters in dozens of communities said no to data centers. Google made it harder for schools to say no to AI by tying it to Search. And Anthropic's IPO filing spells out what investors are being asked to say yes to.
Each story comes back to the same question. As AI becomes part of the infrastructure we all depend on, who gets a say?
As always, my email is hello@wiobyrne.com.
Who Pays for the Cloud?
In this week’s midterm elections, data centers were on the ballot, sometimes literally. How much they shaped the results is hard to say. But I think it’s a sign that people are paying closer attention to the physical side of AI than many expected.
Data centers were the subject of 45 local ballot measures. As of Wednesday morning, 33 had passed in ways that limit new construction. In an NBC News exit poll, 76% of voters said they didn’t want a data center in their community. Data centers also came up in larger races, and almost every candidate had to take a position.
That’s a rough result for the AI industry’s most aggressive wing, which spent the past year looking ready to buy its way through these elections. Instead, its political groups pulled back as anger over data centers grew.
That spending was the visible part of the campaign. Last week, researchers at the University of Pittsburgh’s CTRL lab released a report on the less-visible part: a playbook for making data centers appear more popular than they are. The industry’s main trade group, the Data Center Coalition, set up friendly-sounding local groups like Connected Ohio and Texas Connects to run ads. One AI-industry nonprofit paid for ads to build an email list, then called it a grassroots movement. In Virginia, a developer offered $100 Amazon gift cards to residents who wrote the county in support. This is called astroturfing: fake grassroots support. Now some of AI’s biggest players have teamed up with construction unions to fight local bans.
The pitch usually focuses on jobs and taxes, often framed as money for schools. The numbers rarely add up. Meta pledged a grant to schools near its Ohio data center without specifying the amount, while its tax break there ran into the seven figures. In Cedar Rapids, Iowa, two data center projects worth $1.3 billion combined are required to create just 61 permanent jobs. And most states that hand out these tax breaks won’t say which companies get them.
So why is the industry fighting so hard? Follow the money. Alphabet, Amazon, Meta, and Microsoft plan to spend more than
Why this matters: We talk about “the cloud” as it floats above us. It doesn’t. It sits on land, drinks power and water, runs on borrowed money, and learns from our words. This week, people voted on the part of AI they can see. So the next time someone says AI lives in the cloud, ask three questions: Where’s the building? Who’s paying for it?
No AI, No Search
On November 3, Google announced that its AI tools for K–12 student accounts, including Gemini and Gemini in Classroom, will now be off by default until a school turns them on. That sounds like more control for schools. The catch is that Google Search is off by default too.
Schools can enable Search without enabling Gemini. But Search now comes with built-in AI-generated summaries. In practice, a school that wants no AI for students ends up with no Google Search either. Google’s vice president for education told The New York Times that “Search and AI are deeply integrated.”
This matters most for districts that have pressed pause. New York City and Los Angeles both have temporary bans on generative AI in elementary and middle schools. New York Mayor Zohran Mamdani said Google is “trying to use its monopoly power to force AI into our public schools.” Google says the change gives schools more choice, not less.
Why this matters: For years, good research started with good searching. Write a clear query. Read past the first result. Check the source. Now, the first thing students may see is an AI summary, and schools can’t easily turn it off. Educators who want to keep AI out of the classroom may find it’s not their call. The question is no longer whether students will meet AI, but whether we’ve taught them what to do when they do.
Read the Fine Print
On October 26, Anthropic, the company behind Claude, filed to go public. That means ordinary investors may soon be able to buy shares. It also means Anthropic had to disclose things it doesn't usually talk about.
The numbers are staggering. Anthropic reported $16.3 billion in revenue for the first half of 2026, alongside an $80.7 billion loss, mostly from accounting adjustments. Its latest filing puts future infrastructure commitments near $900 billion. That's a lot of money tied to the data centers communities were voting on this week.
But the fine print is more interesting.
- It needs a lot of data centers. Anthropic's original filing disclosed about $417 billion in compute commitments, including cloud services, chips, and data centers. Its updated filing puts infrastructure commitments at nearly $900 billion. The company also warns that hotter temperatures could increase the power and water needed to cool its equipment.
- Its models don't always do what they're told. Anthropic reports that models have circumvented security restrictions, given themselves permissions they weren't supposed to have, and misrepresented whether they completed tasks.
- It warns about the worst case. The filing acknowledges that advanced AI could pose catastrophic or existential risks to humanity. Yet the founders' letter at the front promises an "age of miracles." Both claims appear in the same document.
- A few people keep control. A special class of stock gives Anthropic's seven cofounders 50.1% of the voting power. A separate trust elects a majority of the board. Investors can buy shares, but that doesn't mean they'll get to steer the company.
- It's fighting the government. After Anthropic refused to allow its models to be used for mass domestic surveillance or fully autonomous weapons, the Pentagon labeled it a supply-chain risk, and the President ordered federal agencies to stop using its technology. Anthropic warns that the dispute could cost it substantial revenue. The legal fight continues.
Why this matters: A prospectus is one of the few times a company is legally required to show its work. The founders' letter tells you what the company hopes to accomplish. The risk factors tell you what could go wrong, including problems the company has already encountered. Reading the two side by side is a useful media-literacy exercise, whether you're considering buying stock or choosing a tool for your school. What does the company promise up front, and what does it admit in the fine print?
The Understory
In 1933, when Franklin Roosevelt signed the law creating the Tennessee Valley Authority (TVA), most rural families in the Tennessee Valley still lived without electricity. Cities had electric lights, refrigerators, and factories powered by machines. Out in the countryside, people still relied on kerosene lamps, wood stoves, and water carried by hand.
The problem wasn't that Americans didn't know how to generate electricity. They did. The problem was getting it to people who couldn't make a power company much money. Running lines across miles of farmland to reach a handful of houses was expensive. Private utilities had little incentive to do it.
The TVA offered a different approach. The federal government would build dams, generate electricity, and use the power to support a region that private investment had largely passed over. It was controversial from the beginning. Utility companies challenged the government's involvement, arguing that it competed unfairly with private enterprise. The courts eventually upheld the TVA's authority.
Electricity reached more communities. Farms and homes changed. But the dams also flooded land, displaced families, and altered rivers. The benefits were real, and so were the costs. Public ownership hadn't made those tradeoffs disappear.
What changed was the question Americans were willing to ask. Electricity had been treated largely as a business: companies built where they expected to earn a return. The TVA challenged the idea that profitability alone should determine who received something increasingly necessary for daily life.
Nearly a century later, the TVA is part of the AI story. In November 2024, its board approved 150 megawatts for xAI's Colossus supercomputer in Memphis, despite concerns from residents and environmental groups. This February, it approved another 150.
The agency created to bring electricity to people the market had passed over now helps power the AI boom. The question isn't whether that power should be used for AI. It's who gets to make that decision, who benefits, and who lives with the costs.
See you next week. As always, my email is hello@wiobyrne.com.
Follow the ideas
This is a living edition of Digitally Literate: a weekly newsletter and connected public notebook. I return to questions raised here as related notes are published and updated.
Related Evergreens
- What Can People Do About AI? — four questions for locating decisions, evidence, access, and accountability.
Previous: Show Your Work · Full archive
More about my writing, teaching, and research at wiobyrne.com.