AI is losing the popularity contest. And it could not be happening at a worse time.

CNBC reported this week on a collision course that should keep every AI executive awake: the American public is turning against artificial intelligence at the exact moment the industry needs public investors to write the biggest checks in IPO history.

The numbers are ugly. A March NBC News survey found 57% of registered voters believe the risks of AI outweigh the benefits. Quinnipiac reported 55% expect AI will do more harm than good in their daily lives. Pew found a majority of Americans are more concerned than excited about increased AI use. Three major polls. Same conclusion. The vibes have shifted.

The Firebomb Made It Personal

The latest signal was extreme. A 20-year-old Texas man allegedly threw a lit Molotov cocktail at Sam Altman's San Francisco home and threatened to burn down OpenAI headquarters. The DA said the crime was motivated by hatred of AI technology. That is not a policy debate anymore. That is someone trying to set a CEO on fire because of what his company builds.

Altman acknowledged "great anxiety about AI" and called for de-escalation. He floated a public wealth fund, a four-day workweek, and payroll tax changes. All reasonable ideas. All reactive. The horse left the barn a while ago.

The Data Center Revolt

Then there is the infrastructure problem. Big Tech has committed roughly $700 billion this year to build out AI data centers across the US. The smarter the models get, the more compute they need, and the more land and power they consume. Local communities have noticed.

Data Center Watch found that in 2025, at least $156 billion in data center projects were blocked or delayed due to local opposition and litigation. This week, Maine passed a bill creating the first statewide data center ban. Voters in Lester, Missouri, threw out several city council members over support for a proposed facility. This is not a fringe movement. It is organized, it is growing, and it is winning.

The IPO Catch-22

This is where it gets genuinely dangerous for the industry. OpenAI and Anthropic are both heading toward IPOs. OpenAI has specifically said it wants to reserve shares for individual investors. CFO Sarah Friar compared it to SpaceX: "Everybody wants to own part of a rocket company. I hope everyone wants to own part of ChatGPT."

But "everyone" includes the 57% of voters who think AI is a net negative. It includes the communities fighting data centers. It includes the people who watched someone try to firebomb the CEO and thought: I understand the anger, even if I would never do that.

OpenAI is valued at $852 billion privately. It burns $85 billion a year. Its business model depends on continued data center expansion in communities that increasingly do not want them, selling subscriptions to consumers who increasingly do not trust the product, and going public to investors who are hearing "AI" and thinking "risk" instead of "opportunity."

The Anthropic Paradox

Anthropic has a different version of this problem. Dario Amodei has been loudly warning about the risks of large-scale disruption from AI. That is good for credibility. It is less good for an IPO roadshow. "Buy our stock, but also, our technology might disrupt half of all white-collar jobs" is not exactly a pitch that screams consumer confidence.

The AI industry has spent two years telling the public that its technology will change everything. The public believed them. Now they are scared. And scared people do not buy IPOs. They vote for politicians who promise to regulate the thing that scares them.

AI is about to have its first true popularity test. It is going to be a stock ticker, not a poll question. And the numbers say the crowd is not friendly.

First reported by CNBC.