AI Spending Doubts Hit Markets as Anthropic Cuts China

Anthropic told investors at a meeting in San Francisco last week that it had cut off paying customers responsible for between 5% and 7% of its revenue in the first nine months of the year. The company alleges Chinese AI labs used those accounts to mine its Claude model. It’s a rare case of an American technology firm giving up revenue for security reasons. The move came in a week when markets began asking tougher questions about what all the AI spending is buying.

The Security Dimension

Anthropic said the Chinese labs DeepSeek, Moonshot and MiniMax generated more than 16 million interactions with Claude through roughly 24,000 fake accounts in February, violating its terms of service. The practice, known as distillation, amounts in the company’s view to a loss of intellectual property with national security implications.

Earlier this month, Anthropic introduced access restrictions, blocked proxies and banned tens of thousands of accounts. The company said those steps cut off 90% to 99% of the data being siphoned and effectively stopped Chinese developers from distilling its models. Enforcing rules like these amounts to border control in digital form: terms of service mean little without the ability to police who gets in.

The commercial pressure is plain. Anthropic is nearing an IPO and faces investor concerns about month-to-month growth deceleration. Revenue reached $4.7 billion in the first quarter, $11.6 billion in the second and $16.8 billion in the third. Some investors have discussed an IPO valuation near $1.5 trillion, down from the roughly $2 trillion once expected. Cheaper, open-source Chinese models remain a worry.

The Fiscal Picture

Broader markets spent Thursday weighing the cost of the AI build-out. Chipmakers plunged nearly 7%, sending the Nasdaq 100 to its worst day since June, while the S&P 500 fell 1.2%. Meta Platforms’ plans to raise capital spending revived concerns that technology companies would issue more debt to fund AI infrastructure.

In Europe, Paris’s CAC 40 fell 1.4% and Frankfurt’s DAX 40 lost 1.1%. London’s FTSE 100 finished down 27.85 points, or 0.3%, at 10,725.03, while the FTSE 250 fell 0.9%. Meta dropped 5.6% despite better-than-expected third-quarter revenue. Chief financial officer Susan Li said the computing intensity of its AI agent Muse meant that expanding its user base would require significant investment. She also expected total serving costs to rise.

James McCann of Edward Jones said investors wanted stronger evidence of returns from AI investment and questioned whether spending could continue amid higher rates. Trade Nation analyst David Morrison put it bluntly: “The future commitments are enormous. Yet there’s little clarity over when, or even if, there will be a return on planned expenditure.” He called the reaction to company updates “concerning.”

There’s disagreement on the risk. Blackstone President Jon Gray said investors needn’t fear an AI bubble, days after billionaire Ken Griffin said the industry was “almost certainly” headed in that direction. “With AI, you’re building capacity, and you’re hoping you can keep up with the demand,” Gray told Bloomberg Television. Nvidia chief executive Jensen Huang told CNBC demand for neoclouds was “off the charts,” but capital was scarce and held back growth.

Energy and Rates

Brent crude stood at $111.18 a barrel at the London close, up from $109.14 a day earlier. The New York Times reported, citing sources, that the United States and Israel had plans for renewed strikes against Iran after next month’s U.S. midterm elections. Saudi oil company Aramco told remaining staff to evacuate its headquarters Thursday, two sources told AFP, after weeks of Houthi attacks. An attack on Dammam airport killed two people and injured eight.

Money markets no longer fully priced in a Federal Reserve rate increase by the end of 2026, and Treasuries posted their biggest jump in more than a year. The U.S. 10-year yield stood at 5.02%, down from 5.11%. Ian Lyngen of BMO Capital Markets said the tech correction led investors to reconsider whether a market-induced tightening of financial conditions could justify a less restrictive course for monetary policy.

Chinese stocks entered a bear market. They had lost $2.4 trillion since a peak in May, amid a prolonged consumption slump and doubts about AI companies’ profitability.

Corporate Winners and Losers

London Stock Exchange Group rose 7.7% after raising its annual guidance and confirming its $3.54 billion acquisition of Octus. Rentokil Initial gained 8.6% after announcing a $200 million share buyback. Inchcape fell 9.3% after cutting its 2026 margin guidance to around 5.5% from 6.0%, citing a “challenging situation in Australia.” J Sainsbury fell 2.2% after like-for-like sales growth excluding fuel of 1.2% missed consensus of 1.9%. Falling metal prices dragged mining shares and the FTSE 100.

A Salt Security survey of more than 380 security and technology professionals examined the governance of AI itself. It found 86% of their organizations had at least one AI agent operating. Only 18% said they could automatically stop an agent behaving unexpectedly, and 13% could trace a complete path from instruction to outcome across every agent involved.

Why This Matters:

Three threads ran through Thursday’s trading. First, intellectual property is now a security question: Anthropic’s decision to forgo up to 7% of revenue shows guarding strategic technology has a price, and European firms and governments will face the same choice. Second, financing AI infrastructure is shifting from a growth story to a debt-and-returns story. Rates above 5% on the 10-year Treasury raise the cost of capital that Europe’s tech sector must compete against. Third, only 18% of firms say they can automatically halt a misbehaving AI agent, a gap regulators will be tempted to address with rules. Brussels faces a competitiveness test: can it demand accountability without adding burdens that push investment elsewhere?