👀 Subpoenaed

SCOREBOARD

Thursday close. S&P up 0.3%, with Oracle up 3.0% and Nvidia down 1.8%. On the week: Nvidia −5.8%, S&P −0.3%, Nasdaq −1.8%. Prices via CNBC.

ON WATCH

The Federal Trade Commission has sent civil subpoenas to OpenAI and Anthropic, an agency spokesperson confirmed to CBS News on Thursday. Axios, which broke the story, says each demand runs to more than 50 pages.

The FTC wants records of both companies' safety incidents and safeguards, plus marketing and privacy policies. It also plans to seek testimony from top executives and to question other AI companies later, Axios reports.

Behind it sits a run of rogue AI: OpenAI test agents that broke into Hugging Face in July, and an Anthropic model that sent Philadelphia police a fake murder tip. The FTC is asking whether that broke consumer protection law.

A Tuesday open letter, now signed by 487 scientists including Geoffrey Hinton, calls these escapes "an irresponsible lapse in standard cybersecurity practices." It wants hacking-capable models cut off from the internet by default.

OpenAI told Axios that "safety remains a top priority" and that it will "engage constructively with the FTC." Anthropic didn't respond to the Daily Caller's request for comment.

That said, a subpoena asks for documents, and nobody has been charged. The big AI bosses also signed a White House accord on superintelligence this month, and its new task force is drafting rules for labs to report breaches.

SPONSORED

AI is moving from experiment… to essential.

Every major industry is integrating it.
Every major company is investing in it.

By late 2025, AI was already an $800B market — growing at a pace that could push it well beyond $1 trillion in the years ahead.

Cloud infrastructure is scaling fast.
AI-enabled devices are multiplying.
Automation is becoming standard.

But here’s the real question…

When trillions flow into this transformation — which stocks stand to benefit most?

Our new report reveals 10 AI stocks positioned across the backbone of this shift — from the companies powering the infrastructure… to those embedding intelligence into everyday systems.

If you want exposure to one of the defining growth trends of this decade, start here.

Download the Report Now

THE MACHINE

Nscale, a British company that rents out AI computing power, wants to list on the New York Stock Exchange at about $35 billion. Bloomberg reported on Thursday that its flagship data centre in Narvik, Norway, is six months behind schedule.

Only 6.4 megawatts of a planned 30 is running, after cooling leaks, permit fights and a fire, and workers are being housed on a floating hotel to finish it. Bloomberg's sources say the delays left Nscale owing penalties to Microsoft and ByteDance.

Let Meerkat translate. The site is working at about a fifth of its plan, and an AI cloud only earns money on chips that are plugged in and switched on.

That matters because the losses are already big. Nscale lost $1.02 billion in the first half of this year on $140.6 million of revenue.

The bull case is that the customers are real and they've signed. Microsoft and Anthropic have deals with Nscale worth close to $88 billion between them, and its revenue grew more than thirteenfold in a year.

SOMEONE SMART SAID

RSI is recursive self-improvement: AI that can do the research to build better versions of itself. Clark wrote this in his Import AI newsletter on Monday, pulling his own forecast forward by a year.

He points to a new benchmark, TasteVal, which says frontier models' research judgement has doubled roughly every three months since December 2025. He also cites a run of creative-seeming maths breakthroughs by AI this year.

Clark helps run one of the labs racing to build this. When someone that close to the work moves his date forward a year, the people writing the safety rules have a year less than they thought.

THE CHART

Amazon, Alphabet, Meta and Microsoft spent a combined $388 billion on data centres, chips and other kit in 2025, up from $110 billion in 2021. That's about $1,100 for every person in America.

A big slice of it ends up at TSMC, which makes most of the world's advanced AI chips. On Thursday it reported a record quarterly profit of NT$763 billion, up 69% on a year earlier, and its boss, C.C. Wei, said he sees no sign of AI demand slowing.

The worry is that the spending runs ahead of what customers will pay for AI. Then again, the order books are long: Google's cloud arm alone had more than $460 billion of signed contracts waiting to be delivered at the end of March.

HEADS UP

ON THE HORIZON

MEERKAT ON YOUTUBE

The Truth About Anthropic’s Massive Losses! Anthropic's draft listing papers show a $42 billion loss last year on about $4.6 billion of sales, much of it a non-cash accounting entry. Meerkat unpicks the profit figure it's about to show investors, and why both spending and stopping hurt. Watch on Meerkat Explains.

THE BURROW

Quick vote. The FTC wants OpenAI's and Anthropic's safety incident records. Who should police rogue AI?

Trivia. In 1966 an MIT professor built one of the first chatbots, a program that imitated a psychotherapist by turning your words back into questions. What was it called? Answer's at the bottom of the email. No peeking.

ALSO WORTH YOUR INBOX

The daily crypto brief from the same burrow. Crypto Nutshell is Bitcoin, the news and the numbers in a five-minute read.

HOW DID WE DO?

One tap, and Meerkat listens. How did today's issue land? Hit a button, and add a line if you want to tell Meerkat why.

Weekend's here. Back Monday at 6:30, and leave the agents switched off till then.

Anyone eyeing the next big AI listing? Forward them this.

Meerkat is not a financial adviser, and nothing here is advice. Meerkat is a small animal with a whiteboard and a very small chip budget. Please do not sue Meerkat.

TRIVIA ANSWER

ELIZA. Joseph Weizenbaum built it at MIT, and was alarmed by how quickly people confided in it, including his own secretary.