Voters push back on data centers, Apple rations AI, and Uber's four AI rides

Wednesday, November 4, was the day after the US midterm elections, and a lot of the AI news was about AI running into the real world: voters, budgets, habits and the news business.

Data centers, the giant buildings full of computers that AI runs on, turned out to be a real election issue, and they lost almost every local vote they were part of. Apple is reportedly cutting how much its own engineers can spend on AI each day. Uber said that during its busiest week in October, only four rides were booked through an AI agent. Meta bought Conductor, a startup that runs teams of AI coding agents. And Adweek reports that Anthropic, the company behind the Claude chatbot, has signed its first paid licensing deal with a news publisher, the Associated Press.

Here's what happened, in plain English.

Voters push back on data centers

A data center is a very large building packed with computers. It's where the internet lives, and it's where AI models are trained and run. AI companies need a lot of them, and they need a lot of electricity to keep them going. On Tuesday, November 3, voters in several states got a direct say on them.

According to CNBC, citing Ballotpedia, a site that tracks elections, there were 41 local ballot measures in 10 states that would restrict data center construction in different ways. Ohio alone had 21 of them, more than half. Communities in California, Florida, Illinois, Kansas, Maine, New Jersey, Nevada and Wisconsin voted on the issue too.

The results, gathered by CNBC on Wednesday, were close to one-sided. Voters sided against data centers in every measure except one. The lone exception was Harwood, North Dakota, where voters rejected an added review of industrial projects, including data centers.

The issue also reached the bigger races. In a CNBC voter poll, 56% of voters said data centers were a crucial or major factor in their decision, and Democrats ranked the issue higher than Republicans did. In Ohio, the governor's race was won by Democrat Amy Acton, a physician, who supports what CNBC describes as a "conditional moratorium" on data centers. In plain terms, that's a pause with conditions: data centers would have to cover 100% of their utility costs and, in her words, "exclusively build with union labor."

That matters because Ohio is where some of the biggest AI building plans are. CNBC says OpenAI announced in August that it had secured 8 gigawatts of computing capacity at the PORTS-Pike Technology Campus, which CNBC calls the largest planned AI data center campus in the world. TechCrunch reports that analysts at Morgan Stanley expect Acton to pose a major risk to Ohio's data center buildout, and that a county where Meta planned a 700,000-square-foot facility has just passed a ban on data centers.

TechCrunch's reading of the day is that anger over data centers is becoming a political force. It also points to an NBC exit poll in which almost 80% of respondents opposed building data centers in their own neighborhoods.

The AI money in the election

There was also a lot of money from the AI world in this election, much of it spent through PACs. A PAC, or political action committee, is a group that raises and spends money to influence elections.

TechCrunch looked at how the AI-focused groups did, and the picture was mixed. Two AI industry PACs, American Mission and Building the Future, together spent $4.8 million backing South Carolina Senate candidate Darline Graham, who lost. American Mission is the Republican spending arm of Leading the Future, a broader group backed by OpenAI president Greg Brockman and the venture capital firm Andreessen Horowitz. Leading the Future's Democratic arm, Think Big, backed Florida Rep. Jared Moskowitz, who won reelection.

Public First, a group backed by Anthropic, saw several Republicans it supported lose, but had more success with the Democrats it backed. On the other side is Guardrails Alliance, which opposes candidates backed by the AI industry. Its political director, Charlie Blaettler, told TechCrunch: "Our goal is to make AI money politically toxic."

TechCrunch is careful to add that it's hard to separate the effect of AI-related spending from what it calls the broader Democratic wave on Tuesday.

What comes next

TechCrunch reports that the results gave Democrats full control of Congress, with a Republican in the White House, and it expects a bigger clash over data centers as a result. Some of that will happen in states and towns. Some may happen in Washington, where Sen. Bernie Sanders and Rep. Alexandria Ocasio-Cortez introduced a bill earlier this year that would put a nationwide moratorium on new data centers until Congress adds guardrails. TechCrunch notes it could face a veto, and that even among Democrats, a blanket freeze is contested.

Apple puts its engineers on an AI budget

The second story is about AI bills, the kind that land on a company's desk.

TechCrunch, citing a Bloomberg report from Wednesday, says Apple is cutting the daily amount its engineering teams can spend on AI.

Some background helps here. When engineers use AI coding tools, every request costs money. The cost is counted in tokens, the small chunks of text that AI models read and write, and AI companies charge by the token. A long, complicated coding session can use a lot of them.

According to Bloomberg, Apple lowered the daily AI spending cap for its engineering teams in October, from $300 to $200. More cuts are reportedly in the works, with some allowances falling to $100 a day. Outside engineering, Apple is reportedly weighing a daily limit as low as $25. TechCrunch says it reached out to Apple for comment.

TechCrunch puts this in a wider frame. It describes a recent frenzy of AI use inside tech companies, which it calls "tokenmaxxing," that turned out to be short-lived once the bills came due. Companies, it says, decided that token rationing might be the more sensible approach.

It's a small story with a big idea inside it. AI tools aren't free to run, even for one of the richest companies in the world, and companies are starting to ask how much value they're getting for what they spend.

Uber: four rides booked by AI agents

Next, a reality check for one of the most talked-about ideas in AI right now.

An AI agent is AI that doesn't just answer questions, but does things for you: booking a ride, ordering dinner, filling in a form. Many tech companies are betting that people will soon hand everyday errands like these to a chatbot.

On Uber's third-quarter earnings call on Wednesday, the company said that only four rides were ordered through an AI agent during its busiest week in October, the week of Halloween, as TechCrunch reports. More people used AI agents to order food through Uber Eats, but the numbers were still small. According to a call transcript cited by TechCrunch, chief financial officer Balaji Krishnamurthy told analysts that Uber and Uber Eats handled around 320 million trips and orders that week, and only about 500 Uber Eats orders were placed by an AI agent.

CEO Dara Khosrowshahi was frank about it: "It is very, very early and very limited traction that we are seeing so far." He also noted that booking a ride in the Uber app takes about 50 seconds on average, and reordering a favorite meal about 90 seconds. TechCrunch's conclusion is that an AI agent isn't necessarily saving anyone time.

Uber isn't writing agents off. Khosrowshahi said there are around eight major AI assistants today and he expects more than 10 over time, that "every single AI system company is talking to us," and that he thinks agents will be a net positive for demand. Uber isn't alone in the low numbers either: TechCrunch notes that DoorDash said last week it had received only a few hundred orders from AI agents.

Meta buys AI coding startup Conductor

Now, a deal in the fast-moving world of AI coding.

Meta has acquired Conductor, a startup whose software lets people spin up software projects by sending several AI agents to work in parallel workspaces. A simple way to picture it: a manager for a team of AI coders, where each one works on its own piece of the project at the same time. Conductor doesn't build those agents itself. They come from major labs such as Anthropic and OpenAI.

Conductor's founder and CEO, Charlie Holtz, announced the news on Wednesday: "I have big news: Conductor is joining Meta." He also wrote that the team has "big, crazy plans for the next interface," and teased something new for customers early next year. Meta confirmed the acquisition to TechCrunch but didn't share the price or other details.

If you use Conductor, there's one date to note. Its core features keep working until early next year, but its cloud service shuts down on November 20, after which the company will begin deleting user cloud data.

TechCrunch frames the deal as part of Meta's push to catch up after lagging behind in AI. Meta launched its own coding agent, Muse Code, in August.

Anthropic's first news licensing deal

And finally, a deal between AI and the news business.

Adweek reports that Anthropic has struck a content and data licensing deal with the Associated Press, one of the world's largest news agencies, covering election coverage. A licensing deal means the AI company pays to use a publisher's reporting, so its chatbot can draw on that reporting when people ask about the news.

Adweek says the deal was struck just before the midterm elections, citing two people familiar with the matter. Anthropic did not respond to Adweek's request for comment, the AP did not comment on the record, and neither side disclosed the financial terms or how long the deal runs.

According to Adweek, it's the first time Anthropic has paid a digital news publisher to license its content. That sets it apart from OpenAI, which Adweek says has negotiated more than a dozen content licensing deals with major publishers, including Axel Springer, News Corp and The Atlantic.

Adweek's own view is that this limited deal doesn't mean Anthropic has changed its overall licensing strategy. Still, it's worth watching. Elections are exactly the kind of moment when people want fast, reliable information, and more of them are now asking chatbots for it.

The big picture

If there's one thread through Wednesday's news, it's AI running into the real world and its limits.

Voters in dozens of communities said no to the buildings AI depends on, and many of them said it mattered to how they voted. Apple, with all its resources, is reportedly putting a daily price limit on AI for its own engineers. Uber, with hundreds of millions of trips and orders in a single week, counted four rides booked by AI agents. And an AI company that had avoided paying for news has, according to Adweek, started doing it, at least for the election.

None of this means AI is slowing down. Meta is still buying startups, and Uber says every AI company is talking to it. But it does show that the next phase of AI won't just be decided in labs. It'll be decided by voters, budgets, everyday habits and deals with the people who make the news.