👀 Two more quit DeepMind

SCOREBOARD

Monday close. S&P down 0.5%, with Nvidia leading the fall as the chip stocks took the slowdown talk hardest. On the week: Nvidia −8.42%, S&P −1.28%, Nasdaq −1.21%.

ON WATCH

Two more researchers have quit Google DeepMind's AI-safety team in the space of a week, and both walked out talking about human extinction. Bilal Chughtai, who worked on AGI safety and alignment, wrote on X that he "earnestly believe[s] that AI has the potential to kill us all."

Josh Engels, from the same safety group, left for the evaluation nonprofit METR after what he called three weeks of contemplation. He warned of a "terrifying chance that AI systems cause immense harm in the next five years," and said he turned down offers from other big labs to do it.

This is starting to look like a pattern. Two weeks ago Anthropic's Jacob Coxon quit with almost the same message, and each departure lands the same way: the people hired to check the brakes keep resigning to shout that the brakes don't work.

To be fair, two exits from a safety team of hundreds are a long way from a verdict, and a sincere belief isn't evidence. DeepMind still runs one of the largest safety operations in the field, and Meta's chief AI scientist thinks the whole panic is recycled nonsense, as you'll see below.

SPONSORED

You've worked hard to fund your portfolio — your investment strategy should work just as hard to maximize your after-tax returns.

On September 17, join Range's CFPs and CPAs live for the practical moves that put more of your returns back in your pocket.

What we'll cover:

Range is all-in-one AI wealth management — tax, investments, retirement, and estate in one place. Bring your questions for the live Q&A. Free to attend, and seats are limited.

Reserve your spot today

^{This webinar is for informational purposes only and does not constitute investment advice or a recommendation to buy, hold, or sell any security. Forward-looking statements involve risks and uncertainties. Past performance is not indicative of future results. Range defines "high earners" as households with income over $300k.}

THE MACHINE

On Monday the warning showed up where it hurts. The chip stocks took the hit while the rest of Big Tech shrugged: Nvidia fell 3.4% and the Philadelphia Semiconductor Index dropped almost 6%, its worst day in months.

Microsoft, Alphabet and Meta all rose the same day. Let Meerkat translate: the market has stopped buying "AI" as one thing, and started splitting it into the firms that must keep buying chips to stay in the race and the platform owners who could stop building tomorrow and still bank the returns.

"They'll just all stop building data centers and just digest what they have," is how D.A. Davidson's Gil Luria put it. Nvidia is the name being asked the hardest question, because its boom depends on everyone else spending freely.

The case against the panic is that nobody has actually cut anything yet. As Luria also noted, no lab has announced a real moratorium, so Monday was a bet on words, not on a change in the build.

SOMEONE SMART SAID

LeCun is one of the three "godfathers of AI" who share computing's top prize, and he still runs Meta's AI research. So when he tells everyone to laugh at the slowdown warnings, it carries further than a random reply. On X his jab was blunt: Amodei called GPT-2 too dangerous to release back in 2019, the world kept turning, and the same script is running again now.

He's no neutral judge. Meta profits when AI moves fast, and open models are his life's cause. But he's also the field's loudest voice insisting the doom numbers are invented. The same expertise quitting DeepMind in fear is, one office over, telling everyone to calm down.

THE CHART

While the labs argued about doom, borrowing costs quietly climbed. The 30-year Treasury yield sits around 5.35%, near its highest in a generation, and on Monday the 10-year briefly touched 5% for the first time since 2023, right before Wednesday's near-certain Fed hike. Higher long rates lean on the AI trade twice over: they make far-off AI profits worth less today, and they make the debt-funded data-centre build costlier to finance.

The counter-case is simple. Rates this high aren't automatically a crisis, the economy has kept growing through them, and a single hike is already priced in. The worry is the direction, not the level.

HEADS UP

ON THE HORIZON

MEERKAT ON YOUTUBE

The Car Repossession Crisis is Getting WORSE in 2026. Missed-payment rates are climbing back toward the levels of the last crunch, and Meerkat walks through who's getting caught and why. Watch on Meerkat Explains Finance.

THE BURROW

Quick vote. Two more AI-safety researchers just quit the biggest labs, warning the tech could kill us all. What's your read?

Trivia. Yann LeCun, who told everyone to laugh at the doomers, shares computing's Turing Award with two other "godfathers of AI." Name either. Answer's at the bottom of the email. No peeking.

ALSO WORTH YOUR INBOX

Meerkat's sister read. Crypto Nutshell does for crypto what Meerkat does for AI: the money, explained without the hype.

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.

The Fed decides tomorrow at 2. Meerkat will be watching the dot plot, not the drama.

Got a mate who reckons the people building AI never lose sleep? Forward them this.

Meerkat is not telling you to buy anything, sell anything, or short anything. Meerkat is not a financial adviser. Meerkat is a small animal with a whiteboard and a chip stock it will not try to catch. Please do not sue Meerkat.

TRIVIA ANSWER

Geoffrey Hinton and Yoshua Bengio. The three shared the 2018 Turing Award for the deep-learning work that made today's AI boom possible.