AI Industry Daily Briefing — November 5, 2026

The Executive Read

For two years the AI industry has mostly been asked how fast it can build. Wednesday’s documents show the people on the other side of the transaction setting terms. In NBC News’s exit poll from Tuesday’s midterms, 76% of voters said they oppose building data centers in their communities, and 57% said AI will do more to hurt society than help it. Bloomberg reported, as relayed by TechCrunch, that Apple has lowered the daily AI spending cap for its engineering teams from $300 to $200, with further cuts planned. Uber’s executives told investors that AI agents placed four rides in its busiest week of October. And Google, one day after announcing that schools declining its AI tools would lose Search, said it will turn AI off by default instead. None of this shows demand falling. Anthropic’s compute spending and AMD’s forecast, both covered yesterday, point the other way. What it shows is that the buyers, the neighbors and the voters now have leverage, and that several companies adjusted within hours or months of being told so. Meanwhile Meta bought a small coding-agent startup, and the first pure-play humanoid-robot company began trading, on a revenue base that is tiny next to its losses.

Top AI Headlines

Voters Say No to Data Centers, in NBC’s Exit Poll

What happened. NBC News’s 2026 exit poll, conducted by SSRS, asked voters about AI. Fifty-seven percent said AI will do more to hurt society than help it; 24% said more to help; 18% said neither. Seventy-six percent said they oppose building data centers in their community; 24% favor it. Fifty-six percent called AI and data centers a crucial or major factor in their congressional vote. On which party is better able to handle AI, 30% chose Democrats, 27% Republicans and 43% neither. NBC notes these are exit-poll responses, not vote results. TechCrunch reports that Democrats won control of Congress, that candidates in several states won after opposing rapid data center buildout, and that a county in Ohio where Meta had planned a 700,000-square-foot facility passed a ban on large data centers, citing the Dayton Daily News. TechCrunch also reports that Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez introduced a bill for a nationwide moratorium on new data centers. We did not open the bill text.

Why it matters. TechCrunch’s own write-up rounds the poll up to “almost 80%” and “almost 60%”; we use NBC’s figures. Opposition was partisan at the margin: among those who oppose local data centers, 64% voted Democratic and 34% Republican, per NBC. But opposition was majority opinion overall. The poll does not say why voters object. TechCrunch lists electricity costs, tax breaks and gas turbines as grievances, but those are the reporter’s framing, not poll findings.

Business implication. Power is already the slowest part of building a data center, as yesterday’s report on converted jet engines showed. Local approval now joins it. TechCrunch reports analysts expect moratoriums and rules making data centers pay for their own grid upgrades, and a Goldman Sachs forecast of delays in Georgia and Texas. We have not read either analysis; treat them as forecasts.

Sources: NBC News, 2026 exit poll results; NBC News, exit poll analysis; TechCrunch, “After the midterms, a bigger clash over data centers could be coming” (November 4).

Google Backs Off Its Plan to Cut Schools’ Search Access

What happened. On November 3, Google said schools using its K-12 software would lose Search unless they opted in to its AI tools. Chris Phillips, a Google vice president for education, told the New York Times that “Search and AI are deeply integrated.” Backlash followed. TechCrunch reports that Senators Bernie Sanders and Richard Blumenthal, former FTC chair Lina Khan and New York City Mayor Zohran Mamdani criticized the plan; Sanders called it “extortion.” On November 4, the Google News account said on X that Google had listened to educators and would turn AI products off by default. According to TechCrunch’s account, Gemini, Gemini Notebook and Search would all be off by default and schools could switch each on individually. TechCrunch says it is not clear whether the original Search requirement is fully withdrawn.

Correction to yesterday’s edition. Our November 4 edition said Google “plans to turn AI tools off by default” and that schools restricting AI entirely would lose Search, based on TechCrunch’s first report. That described the November 3 plan. It has since been revised, and we have not read Google’s own announcement or the post’s full wording. Nothing in yesterday’s text is withdrawn as inaccurate to its source, but it is now out of date.

Why it matters. Google reversed a policy within about a day, which is unusual for a company of its size. Google says it is also developing a “Guided Learning” tool for student Search accounts that walks students through problems instead of giving answers. That is a plan, not a release.

Business implication. Bundling a must-have product with an optional one invites exactly this reaction, and administrators now know they can push back. Other vendors selling AI into schools, including DuckDuckGo, which TechCrunch reports is pitching itself to schools, will use the opening.

Sources: Google News on X, clarification (November 4); TechCrunch, “Google faces backlash for plan that would turn off Search for schools that restrict AI” (November 4); New York Times, original report (November 3), not opened by us.

Apple Is Reportedly Rationing Its Own Employees’ AI Use

What happened. Bloomberg reported on November 4 that Apple has cut internal AI usage budgets. As relayed by TechCrunch, which had not received a response from Apple, the daily spending cap for engineering teams was lowered in October from $300 to $200, some engineering allowances are set to fall to $100 a day, and Apple is considering a limit as low as $25 a day for staff outside engineering. The reporting is Bloomberg’s; Apple has not confirmed it in anything we found. We could not open Bloomberg’s article and rely on TechCrunch’s summary.

Why it matters. Many companies pushed staff to use AI tools heavily this year, a habit TechCrunch calls “tokenmaxxing.” A token is a chunk of text, about three-quarters of a word, and AI services charge per token. If the report is right, one of the world’s richest companies is treating the bill as a cost to manage. The reports do not say what Apple measured about the benefit, or whether the cuts reflect price, productivity or both.

Business implication. Finance teams will ask what each dollar of AI spending returns, and vendors billing by usage are exposed if customers set caps. This pairs with Anthropic’s reported compute costs yesterday: the expense is real at both ends. This is informational, not investment advice.

Sources: Bloomberg, “Apple cuts internal AI usage budgets” (November 4), reporting is Bloomberg’s, paywalled; TechCrunch, “Apple is reportedly cutting budgets for its internal AI usage” (November 4).

Uber: Four Rides Through AI Agents in Its Busiest Week

What happened. On Uber’s third-quarter earnings call on November 4, as reported by TechCrunch’s Sarah Perez, executives said only four rides were ordered through an AI agent during the week of Halloween, one of its busiest of October. About 500 Uber Eats orders came through agents. Chief Financial Officer Balaji Krishnamurthy put total activity that week at roughly 320 million trips and orders. CEO Dara Khosrowshahi said, “It is very, very early and very limited traction that we are seeing so far,” and noted that booking a ride takes about 50 seconds in the app, so agents may not save much time. He also said “every single AI system company is talking to us.” TechCrunch adds that DoorDash has reported only a few hundred agent-placed orders, and that Meta said its Muse app passed 10 million users.

Why it matters. An AI agent is software that completes a multistep task, such as ordering a ride, on a person’s behalf. The pitch is that people will stop opening apps. Uber’s numbers say that has barely begun, at least for rides and delivery. Four out of hundreds of millions is a rounding error, though the figures are one week at one company.

Business implication. Companies building agent features for consumer purchases should expect low early volume where the existing app is already fast. Uber is positioning itself as one more channel agents can reach, not a casualty of them. We could not open Uber’s investor-relations release and rely on TechCrunch’s account of the call. This is informational, not investment advice.

Sources: TechCrunch, “Almost no one uses AI agents to book their Uber, company says” (November 4); Uber Investor Relations, earnings materials not located by us.

Model and Product Updates

Meta acquires Conductor, an AI coding tool. Axios first reported, and TechCrunch confirmed with Meta, that Meta has acquired Conductor, a startup whose product lets people run several coding agents side by side in separate workspaces. Terms were not disclosed. Conductor’s own announcement says it is “joining Meta” and that a new product is coming. Its timeline: sign-ups closed November 4; the cloud version becomes read-only on November 13, with refunds for active subscribers; the cloud service shuts down on November 20, after which deletion of cloud data begins; the local Mac version keeps running until early 2027, then winds down. Conductor says it has never sold or trained on user data. TechCrunch notes Conductor’s agents came from outside labs such as Anthropic and OpenAI, that Meta launched its own coding agent, Muse Code, in August, and that Meta’s Muse agents reportedly passed 10 million users. Meta’s confirmation to TechCrunch is the only public Meta comment we found. For Conductor’s customers, the practical step is to export data before November 20.

Anthropic expands free research credits. An October 28 post says Anthropic’s AI for Science program, which gives free Claude credits to researchers, now extends to funders and institutions, including the Simons Foundation, Schmidt Sciences, the U.S. National Science Foundation, the Wellcome Sanger Institute and the Francis Crick Institute. The post does not state a credit amount or dollar value, so the scale is unknown. This is a company announcement.

Nothing says its first “AI-native” device arrives in December. TechCrunch reports CEO Carl Pei posted on X that the product launches next month, has no screen and “couldn’t have been an app.” No date, price or specification has been given.

Sources: Conductor, “Conductor is joining Meta”; Axios, Meta acquires Conductor (November 4), not opened by us; TechCrunch, “Meta acquires AI coding startup Conductor” (November 4); Anthropic, “AI for Science expands to foundations and research institutions” (October 28); TechCrunch, Nothing (November 4).

Regulation and Policy Watch

AI money in the midterms: mixed results. TechCrunch’s November 4 review of AI-focused political action committees (PACs, which pool money to back candidates) reports that Leading the Future, a network backed by OpenAI president Greg Brockman and the venture firm Andreessen Horowitz, spent $27.38 million in primaries, including $8 million against New York Assembly member Alex Bores, who lost his congressional bid in June. Its Republican arm, American Mission, and Building the Future together spent $4.8 million backing South Carolina Senate candidate Darline Graham, who lost. Public First, backed by Anthropic, supported several Republicans who lost and did better with Democrats, and gave $3 million to a PAC supporting California state Senator Scott Wiener, who was leading when TechCrunch published. Guardrails Alliance spent more than $1 million each opposing two candidates who lost. TechCrunch says it is hard to separate AI spending from the broader Democratic wave. We did not check FEC filings; the figures are TechCrunch’s.

OpenAI’s policy chief frames AI as a 2028 issue. Chris Lehane, OpenAI’s chief global affairs officer, published an essay titled “2028: The AI Election,” per TechCrunch. It compares AI’s economic effect to the New Deal, calls for a “compelling third path” so people share in AI’s benefits, and lists five priorities including worker protections, education, safety and data center development. TechCrunch says Lehane warns that a data center moratorium could cost the U.S. its lead over China. Kate Brennan of the AI Now Institute called the essay self-serving. OpenAI did not respond to TechCrunch before publication. We could not open the essay and rely on TechCrunch’s summary.

Federal procurement: Anthropic’s government discount. In an October 27 post, Anthropic said a new General Services Administration agreement gives government customers a 40% discount on Claude Enterprise, Claude for Government and the Claude API, extending to state, local and tribal governments, with no minimum spend. Anthropic estimates the discounts will save the government $1 billion. It says about 500 federal, state and local organizations use Claude and that Claude for Government is authorized at FedRAMP High, a federal cloud-security level. The savings figure is Anthropic’s estimate. The post is a week old but new to this publication’s record.

United States and European Union. We found no new primary-source filings on federal preemption or the EU AI Act in the last 48 hours, and report none. A House bill reported by one news roundup to revive the Office of Technology Assessment could not be confirmed on Congress.gov, and we do not report it.

Sources: TechCrunch, “AI PACs spent millions trying to shape Congress” (November 4); TechCrunch, “OpenAI is already trying to shape the 2028 election” (November 4); Lehane essay, OpenAI Global Affairs, not opened by us; Anthropic, “Expanding our partnership with GSA” (October 27).

Emerging Startup Radar

Agility Robotics: public listing, small revenue. Agility, maker of the Digit humanoid warehouse robot, is listing on Nasdaq through a merger with a special purpose acquisition company (SPAC, a shell company that raises money first and then buys a business). A filing from the SPAC, Churchill Capital Corp XI, dated October 30 says the combined company was expected to close the deal on November 4 and trade as AGLT on November 5, and that gross capital raised would be about $465 million before expenses: about $164 million from the SPAC’s trust after redemptions, about $201 million from a private placement at $10 a share led by Foxconn, and $100 million already received. TechCrunch’s November 5 article describes Agility as the first pure-play humanoid robot company on the stock market and reports revenue of about $174,000 in the first half of 2026, mostly from related parties, against losses of roughly $362 million. It reports about $300 million in planned contracts from 30 unnamed customers for the next robot, Digit 5, contingent on undisclosed performance targets, with a demonstration expected at CES in January 2027. We did not see a closing filing, so the exact listing time is TechCrunch’s and the SPAC’s expectation. This is informational, not investment advice.

Conductor’s exit. See above: Conductor, founded in 2024 per TechCrunch, is joining Meta and will shut its cloud service on November 20.

Sources: Churchill Capital Corp XI, Form 8-K Exhibit 99.1 (SEC, October 30); TechCrunch, Agility (November 5).

AI Infrastructure and Market Signals

Politics now sits next to power. The NBC poll above is the market signal of the day for builders: 76% opposition at the local level. TechCrunch reports Morgan Stanley analysts flagged Ohio governor-elect Amy Acton as a risk to that state’s buildout, that Texas governor-elect Gina Hinojosa backs a statewide moratorium, and that OpenAI has eyed a $500 billion Ohio data center backed by more than $100 billion in Nvidia financing. We did not open the Reuters report or any filing behind these. TechCrunch also notes Senator Martin Heinrich’s GRID Savings Act would keep grid costs off communities; we did not read it.

The cost of running AI, again. Apple’s reported spending caps and Anthropic’s reported compute-to-revenue ratio, from yesterday, both concern the price of using models. The price of making them is the other half: AMD guided to about $17 billion in revenue next quarter. All three are different companies describing different parts of the same chain.

Sources: NBC News exit poll; TechCrunch, data centers after the midterms (November 4); AMD Q3 2026 release (November 3).

Public Investment Watchlist

Informational only; this is not investment advice.

• Apple (AAPL). Bloomberg reports cuts to internal AI spending caps; Apple has not confirmed. Separately, App Store age-rating changes noted yesterday.
• Uber (UBER). Third-quarter call: four agent-booked rides and about 500 agent-booked Eats orders in a week of roughly 320 million trips and orders, per TechCrunch.
• Alphabet (GOOGL). Revised school policy on AI and Search after criticism; no financial disclosure.
• Meta (META). Acquired Conductor; terms undisclosed. TechCrunch reports Muse agents passed 10 million users.
• Agility Robotics (AGLT). Expected to begin Nasdaq trading November 5 after a SPAC merger, per the SPAC’s October 30 filing; first-half 2026 revenue about $174,000, per TechCrunch.
• AMD (AMD). Covered yesterday: revenue $13.377 billion, Q4 guide about $17 billion.
• Anthropic (proposed listing). S-1 not yet effective; price range and share count not set, per Anthropic’s October 26 post.

Watchlist

• November 12: OpenAI’s planned end of direct model supply to Cursor, which SpaceX acquired, per OpenAI’s August announcement and wire coverage. We did not open OpenAI’s page on the decision and found no later change.
• November 13 and 20: Conductor Cloud goes read-only, then shuts down.
• November 16: Apple age-rating declarations appear in App Store Connect.
• November 18: Ninth Circuit argument in xAI v. Bonta, per a secondary tracker; unconfirmed on the docket.
• December: Nothing’s first AI-native device, no date.
• Anthropic S-1: Price range and share count still unpublished.
• Google: Whether the written policy for schools matches the revised statement.

---

A note on today’s reporting: OpenAI’s newsroom, Reuters and Bloomberg were not reachable by our tools, several lead items rest on TechCrunch’s account of other outlets’ reporting, and we flag each. We dropped an unverified “ChatGPT Images 2.6” release, a reported Anthropic–AP licensing deal, a reported Google lease guarantee for an Anthropic data center, and a Meta bond sale, because we found no primary source for any of them. Yesterday’s note on Google’s school plan is updated above.