Support Us & Upgrade to the Operator Edition

For three years we've guessed at what a frontier AI lab actually earns. As of this morning, we don't have to guess. Anthropic filed its S-1, and the numbers are wilder than the rumors. Meanwhile, the Dutch government, from the land of work-life balance and strong unions, just said AI will let it run with 25,000 fewer civil servants.

This week, we'll deep dive into: the day AI's money question finally got answered on paper.

If you have paid time left on an individual Granola plan, we'll match it with a Wispr Flow subscription that includes Notetaker and dictation, and add bonus time, up to 12 months total. Sign in or create a Wispr account and submit proof of your plan to check eligibility.

Try Notetaker Free

Moonshot AI just released Kimi K3.1 Pro, an open-weight model with a 1M-token context window. It scores 65.7% on Terminal-Bench 4.0, just behind Claude Opus 5.5 at 66.4% and ahead of GPT-6 Astra at 58.2%. Full weights drop by November 4.

The sleeper feature, though, is 3D Make. It turns a text description, sketch, or photo into a model you can print. Watch the 90-second demo, then try it:

💡 Pro tip: Skip the text. Upload a napkin sketch or a photo of something you like, and K3.1 Pro will infer the shape and proportions. It's especially strong on mechanical parts, furniture, and configurable products, so if you sell physical goods, you can go from idea to prototype file in one sitting.

Want the version of this that actually runs in your business?

The 30-Second Play gives you the move. The Operator Brief gives you the full system: prompt libraries, API cost math, and the workflows our subscribers are using to replace entire freelance line items.

This week: How to train your AI platform with skills and then cut down the time you spend prompting and retyping with your other AI tools.

👉 Upgrade to The Operator Brief

💾 SanDisk reports earnings today, and Yahoo Finance argues a strong result could steady a nervous tech market that has stopped trusting the AI giants. A memory chip company’s earnings report is now a referendum on whether the AI trade has legs. The nervous tech market is treating today’s SanDisk report like a canary.

🇨🇳 China's AI bubble didn't wobble. It popped: the STAR50 index is down 40% in under four months, wiping out roughly $480B in market value. Beijing-based fund manager Dong Baozhen said, “The stock market later evolved into a speculative crazy bull. And now we see the bubble bursting.” Western markets are still debating whether the AI bubble will pop. China’s already living in the aftermath.

🏦 Meta launched a $35B stock sale to fund AI, with up to $60B more planned for 2027. Its own chip suppliers, Nvidia, AMD, and Broadcom, may buy up to $5B of it. The AI supply chain itself is now also an investment club.

🚀 SpaceX closed $40B in debt to buy GPUs, and Elon Musk warns the 200+ neoclouds face "guppy-level death and consolidation.” He would know; he just became the biggest fish in the pond. GPUs are expected to arrive in the first half of 2027, so they aren’t a major factor in this quarter, but we’ll see what impact they have for SpaceX in Q1 or Q2 of next year.

🎬 Grok released the first seven minutes of its AI-generated Odyssey, and TechCrunch's verdict was brutal next to Christopher Nolan's box office hit. Some visuals aren’t half bad, even if they still have an uncanny AI feeling, but the main issue is that it feels lifeless. “While a few individual shots might provide a reasonable simulacrum of reality, as a whole, the clip feels completely airless, offering viewers no sense of momentum, no reason to care about anyone or anything on-screen.” The article concludes by stating, “It’s hard to imagine being interested in another seven minutes of this, much less a full movie’s worth.”

Anthropic just filed its S-1 with the SEC and plans to list on the Nasdaq under the ticker ANT.

The prospectus is the first audited look at the economics of a frontier AI lab. It reads like two companies stapled together: a rocket ship and a construction site.

Share count and price range aren't set yet. Fox Business reports the IPO could land as soon as November 19 or 20.

The scary headline is the loss. The real story is the slope. Compute went from 2x revenue to about half of revenue in twelve months. In Q2, revenue grew 1,368% year over year while compute spend grew 300%. Each dollar of Claude usage is starting to pay for itself.

Here's the uncomfortable part:

🔮 Prediction: ANT prices in mid-November, and a strong debut is close to a formality. The reason is scarcity. Until now, public investors could only buy AI through proxies: Nvidia, Broadcom, the hyperscalers. There has never been a listed pure-play frontier model maker. Every growth fund, index provider, and retail investor who wanted direct exposure now has exactly one option.

The bigger effect comes after day one. ANT becomes the reference price for what a frontier model business is worth. Every private lab gets marked against it, from the giants down to the year-old startups raising at $6B valuations. If ANT trades well, expect OpenAI and others to speed up their own filings while the window is open. If it trades poorly, lenders get a public number telling them how much the end customer of all this compute is worth. That would make the neocloud squeeze Elon Musk described this weekend (see below) much worse.

The number to watch isn't revenue. It's the gap between revenue growth and compute growth. That gap is the entire bull case. The S-1 itself says growth will slow, and it has to: roughly 900% net revenue retention cannot last. When growth decelerates but non-cancelable compute contracts keep billing, margins get squeezed from both sides. That makes ANT's first two earnings reports more important than its first trading day.

My read: this is the first AI stock where you can underwrite the business instead of the narrative. That's why it will be hot, and why it will be volatile.

👉 Would you buy ANT on day one? Hit reply with "Yes," "No," or "Wait," plus one line on why.

Follow the Full S-1 Breakdown Here

The Netherlands plans to cut about 25,000 central government jobs over several years in an AI-led efficiency overhaul.

This isn't Silicon Valley, and it isn't Washington. It's a European government in the region best known for writing AI rules, not swinging the AI axe.

State Secretary Eric van der Burg's pitch: a government that is "simpler, smarter and smaller." Parliament has already passed non-binding motions calling for a 20% productivity lift across government using AI.

The surprise isn't that a government is using AI to cut costs. It's which government. For two years, Europe's AI story has been regulation, worker protection, and caution. Now one of the EU's most fiscally careful countries, with a deficit of just 1.8%, is putting a headcount number on AI.

Here's the uncomfortable part:

🔮 Prediction: At least two more EU governments announce AI-linked civil service reductions within 12 months, and almost none will call them cuts.

Here's the reasoning. Every European capital faces the same squeeze: defense budgets rising, populations aging, deficit rules tightening. Notice that police and defense are exempt here. The money freed from administration has an obvious destination.

The Dutch also found the framing that makes this politically survivable. "Buy European" turns a job-cutting program into industrial policy. Ministers can stand in front of voters and talk about sovereignty, not layoffs. Expect that exact language to get copied.

Third, the timeline does the political work. Spread across a decade through hiring freezes and attrition, 25,000 jobs never produces a single dramatic layoff headline. That makes it far harder to organize against than a private-sector cut.

The second-order effect is the one I'd watch most closely. European AI companies have struggled to land anchor customers big enough to compete with US labs. Governments buying "sovereign" AI at national scale could do more for European AI than any regulation ever has.

The real test arrives when citizens start dealing with the new system. If services get faster, this spreads everywhere. If it looks like the benefits scandal again, it stops cold.

👉 If your government announced this tomorrow, would you feel relief or dread? Hit reply with one word. I read every response.

Read More Here

Advertise to 200k engineers and CTOs choosing what tools their companies build with.

👉️ Learn how to sponsor this newsletter

Would love to hear your thoughts! Send me your thoughts by replying to this email (yes, I read them all :)

Until our next AI rendezvous,

Anthony | Founder of Uncover AI