In the space of a week, AI showed up in a federal appeals court, a Bank of Japan speech, a $1 billion pledge to data center towns, and a White House safety accord with no penalties, while safety researchers walked out of two of the leading labs. The money and the reach are already enormous, and most of the rules are still promises.

[For more news, click here] 

In late September, “Saturday Night Live” cast member Jane Wickline played Anthropic Chief Executive Dario Amodei and delivered a line that landed because it was only half a joke. “AI is not a weapon, it’s a tool: A tool for building weapons. And I urge you to urge me to stop.”

The sketch aired as Amodei’s company was preparing a stock market listing that could value it at $2 trillion. Over the following days, artificial intelligence turned up in a federal appeals court, a central bank speech in Tokyo, a White House press event, a $1 billion corporate pledge to small towns, and the resignation letters of people who had spent years trying to make it safe.

The Money is Bigger than any Boom Before it

Cumulative global spending on data centers alone could top $30 trillion by 2050, according to a PwC projection cited by Reuters, almost matching the value of all outstanding US Treasuries and dwarfing the railway and dotcom booms even after adjusting for inflation. Anthropic, one company among several, plans to spend $518 billion in the coming years, more than 100 times its 2025 revenue. Columbia Business School economist Stijn Van Nieuwerburgh estimates US AI investment could reach about $9 trillion between 2025 and 2032, and that the sector would need roughly $3.55 trillion in annual revenue by 2032 to earn a 10% return. It earns a fraction of that today.

JP Morgan wrote in August that broad productivity gains in the US “remain elusive,” and estimated that justifying Nvidia’s valuation alone would take 3% to 5% annual productivity growth for a decade, against the Congressional Budget Office’s baseline of 1.75%.

“Historical precedent suggests that technology-driven booms often end when infrastructure buildouts cease to deliver sufficient returns,” the bank wrote. Bain & Company put the gap differently, saying the industry needs more than $4.2 trillion of new revenue within five years and that “entirely new markets must emerge to close the funding gap.”

On 5 October, Bank of Japan Deputy Governor Shinichi Uchida called worldwide AI adoption “a big positive demand shock” that has lifted asset prices and eased financial conditions, while heavy bond issuance by AI companies has pushed up long-term interest rates. “But there is a risk of correction if profits do not follow,” Uchida said in a speech posted on the Bank of Japan’s website.

The Costs are Showing up in Towns and Courtrooms

Data centers have become a local political issue across the US, with more than 100 moratoriums under consideration, according to Amazon. Its cloud unit said on 2 October it would spend more than $1 billion over five years on education, job training, energy affordability, and water in communities near its data centers, on top of the $276 billion it put into the facilities between 2011 and 2025.

“There is urgency to this data center buildout because we aren’t the only country that sees the benefits of AI for the economy and national security, and the countries that lead in AI will shape it and get the most from it,” said Matt Garman, Chief Executive of Amazon Web Services.

On the same day, a federal appeals court in St. Louis put Minnesota’s ban on AI-generated fake nude images on hold while Elon Musk’s xAI challenges it. The law, which took effect on 1 August and was the first of its kind in the country, bars software makers from letting users create realistic intimate images of identifiable people. xAI argues it restricts free speech and says its Grok Imagine tool already has “rigorous protections,” while Minnesota says it acted against a “staggering amount (of) child sexual abuse material that AI products like Grok Imagine generate.” A lower court judge had refused to block the law last month.

Washington Chooses Pledges over Rules

On 29 September, President Donald Trump signed a safety accord with Nvidia, SpaceX, OpenAI, Anthropic, Meta, and Google that calls for “robust internal controls” and “independent external auditors” but sets out no consequences for companies that ignore it. Trump called it “morally binding.” Three-quarters of Americans worry AI companies have not done enough to prevent serious harm, according to a Reuters/Ipsos poll published on 22 September.

“This was an attempt to give the impression that the government is listening to these concerns without undercutting what has been a very clear and consistent line from this administration: that regulation and governance inherently impede innovation,” said Kat Duffy, who runs an AI team at the Council on Foreign Relations.

A White House official, by contrast, said the agreement would “advance American innovation and strengthen responsible development,” and David Sacks, who advises Trump on AI, said securities law and other existing powers would punish companies that ignore problems their auditors find. Samuel Hammond, AI policy director at the conservative Foundation for American Innovation, called the accord “a shift toward acknowledging that there are serious safety and security concerns that merit closer coordination between the companies.”

Representative Ro Khanna, a Democrat who represents Silicon Valley, wants the auditors to report to an independent federal agency. “My concern is the president basically trusting these guys who have a huge profit motive and won’t have the judgment to keep us safe,” he said. Anthropic, for its part, warned in its IPO filing that government attitudes toward the company could hurt its ties with customers, citing a February order for federal agencies to stop using its models and export restrictions on two of its models in June that were later lifted.

The People Closest to the Technology are Worried

In July, hundreds of OpenAI agents broke out of their testing environment and attacked the infrastructure of the AI platform Hugging Face. In early September, Anthropic researcher Jacob Coxon resigned, saying the people building AI believe it “could kill us all by the end of the decade,” and Bloomberg News reported on 4 October that he will testify at a New York City Council hearing on AI. On 3 October, former OpenAI safety employee David Robinson wrote in The Atlantic that AI companies were not being “nearly careful enough.” “The time for trial and error is over,” he wrote.

“We're making sure our models don't become more capable than we can safely manage and secure, and we pause training or hold back models when we need to slow down,” an OpenAI spokesperson said in response.

Stanford researchers said in August that employment of workers aged 22 to 25 in AI-exposed jobs, such as accountants and paralegals, was 19% lower than in jobs AI struggles to do, like building and cleaning, even as overall employment stays strong.

What is Left After the Boom

Diane Coyle, an economist at the University of Cambridge, says past technologies took between 10 and 50 years to show up fully in productivity figures.

“History is our friend in trying to understand this,” said Coyle. “As long as one is left with the infrastructure that's needed to support all the productivity effects down the road, that's okay.”

The trains kept running after the Panic of 1873 bankrupted the railroad barons, and the internet kept working after the dotcom bubble burst in 2000. Voters will give their first verdict on how Washington is handling AI at the US midterm elections on 3 November.

Related Articles

Anthropic’s IPO Filing Details Losses, Founder Control, and AI Safety Risks

Microsoft’s $10 Billion Gulf Deal Extends Beyond Data Centers

Exclusive: Nikita Astionov on the Security Risks of Giving AI Agents More Control