whatimreading

Wednesday, June 3, 2026 · 146 newsletters

Anthropic Files, Markets Reload

Anthropic IPO · AI capital · Nvidia · AI labor · Pope Leo encyclical · July 4th · Texas Senate · California primaries · NYC · restaurants

Published on Wednesday, June 3, 2026.

Pulled from 146 newsletters in yesterday's inbox. Anthropic confidentially filed its S-1, and the rest of the day, capital markets, Pope Leo's encyclical, the labor data debate, even Nvidia's pivot down market, ended up rotating around what that filing means.

AI Capital: The Anthropic IPO Window Opens

The S-1 was the day's gravitational center. The Information AM led with it, TLDR put it in its top slot, Term Sheet at Fortune ran "Anthropic confidential," Fortune Tech framed it as "Now or never," and Casey Newton at Platformer noted Anthropic found more money on the way to the public markets. Alex Wilhelm's Cautious Optimism asked five questions about what he called AI's biggest IPO. Polymath Investor bundled Anthropic with SpaceX and OpenAI as "the IPOs that could finally break the market," and Techmeme paired the filing with Trump's signed AI executive order extending Project Glasswing.

The capital side of the trade is just as loud. Ben Thompson at Stratechery made the case that "SpaceX ponied up the supply Anthropic needed," and tied the filing to Bloomberg's report that Alphabet plans to raise $80 billion in equity capital purely to feed AI spending. Thompson's companion piece, Mythos, Muse, and the Opportunity Cost of Compute, reads as the theory of the case for why every hyperscaler now needs both a token operation and a capital operation.

The takeaway: the AI cycle has officially moved from venture capital to public capital, and that means the next twelve months will be priced not by model benchmarks but by the cost of equity at three trillion-dollar issuers and one would-be one. The pricing pressure rotates from compute to capital.

AI Infrastructure: Nvidia Goes Down Market

Nvidia did its part to make the capital story tangible. Bloomberg Technology covered "Nvidia laptops are coming," Robinhood's Snacks led with the new PC "superchip," The Code framed it as "Nvidia's new open-weight model," and Superhuman tied the chip news to Trump's AI EO. TLDR called it "Nvidia Introduces First PCs Designed for AI Agents," and Intel's response, "our upcoming AI chip will be cheaper, run cooler than Nvidia, AMD options," landed in the same issue. Sacra ran a deep dive on what it calls "the $10m/yr clawd-ification of Zapier," and Runtime asked whether "Snowflake thinks context is king" now that the data layer is the bottleneck.

The throughline: the moat is moving to the device. After two years of training-side scaling, the spend is migrating to inference at the edge, which is why a $250 million Anthropic deal with a CLI vendor reads as strategy rather than novelty.

AI Labor: The Jobs-pocalypse Gets a Counter

Casey Newton's main piece was the most quoted essay of the day. Newton interviewed labor economist Kathryn Anne Edwards, who argues the AI jobs-pocalypse narrative confuses the Solow paradox with the actual unemployment data. Newton himself notes that what looks like AI displacement is sometimes the politics of an idle class. SmarterX cited Yale's Budget Lab finding no clear employment impact yet from current model adoption.

The contrarian voices got louder too. Noah Smith at Noahpinion asked the heretical question, "how much more software do we really need?", citing Goodhart's Law and the practice of "tokenmaxxing" at Amazon where engineers were encouraged to game leaderboards for token consumption, then quietly canceling Claude Code licenses once the numbers stopped flattering anyone. AMP's "Who is Actually Winning? Claude or ChatGPT?" tracked the same fork from the user side.

The pattern is that the people closest to the spend are getting more skeptical, while the people closest to the capital are getting bolder. Both can be right at once, which is the part the headlines keep missing.

AI Ethics: Pope Leo Becomes Required Reading

This is the trend nobody predicted. Project Liberty's "Five reflections from Pope Leo's letter on AI" took the encyclical seriously as a governance document, tying it to Audrey Tang's pluralist work and to the original Rerum Novarum that gave Leo XIII his name. SmarterX framed the encyclical as "AI's PR Emergency" and paired it with the soaring cost of Claude Code as Silicon Valley's tone-deaf messaging problem. ChinaTalk pushed back, arguing the accelerationists reading Leo as a Butlerian jihadist are misreading him as badly as the doomers, and that "Antiqua et nova" is closer to Christian democracy than to either camp.

The encyclical is doing work the Frontier Risk Report and the Future of Life Institute have not been able to do on their own. When a 2000-year-old institution and a Senate caucus led by Sanders, AOC, Khanna, and Warren cite the same text in the same week, the regulation conversation has a new center of gravity.

Politics: The July 4th Hijack and a Texas Senate Race

Bill Kristol's Bulwark essay was the day's defining piece on the Trump second-term machinery, with the temporarily blocked White House slush fund payments and the staff that fumed and went ballistic with anger when the courts intervened. Joe Perticone at The Bulwark wrote a tough column on the Paxton vs. Platner Texas Senate primary, arguing that Democrats are about to face a character test mirror image of the one Republicans repeatedly failed. JVL's Bulwark essay on Scott Pelley framed the 60 Minutes succession fight as proof that institutional journalism still has a spine where its owners don't.

The Crooked Media feed ran parallel. Matt at What A Day covered Iran's supreme leader "increasingly engaging" with the negotiations, a Russian drone barrage, and AG Todd Blanche confirming dubious fraud investigations, with the column's title nodding to the Trump family real estate beat. Anand Giridharadas used the upcoming America 250 anniversary to argue for a more honest July 4th, in dialogue with Eddie Glaude Jr. and Rebecca Makkai. Paul Krugman explained why he calls economic policy from the second Trump administration "very boring" by design.

The convergence: four different writers on four different platforms all reached for the same metaphor, that the political slush fund is a feature rather than a bug, and that the country has about a month to decide what kind of birthday party it wants.

Primaries: California's Pileup and a Texas Reckoning

California votes today. Democracy Docket's "What we're watching in California's primaries" and Semafor DC's "Primary pileup" both flagged the governor's race as the marquee, with David Callaway arguing the actual fight is for second place behind Katie Porter. Marc Elias at Democracy Docket wrote "They were brought in to kill it," a brutal column on staff purges at the Civil Rights Division. Dan Pfeiffer at The Message Box asked whether Maine's Graham Platner can still win after the week's revelations, and Judd at Popular Information pushed an update on the GOP dirty tricks operation his original reporting exposed.

The interesting tell is how many of these races are being decided on character litmus tests rather than policy. The infrastructure of accountability journalism is healthier than the infrastructure of policy debate, which has consequences neither party has priced in.

NYC: Mamdani Defends His Cops, Prospect Park Mysteries

Mayor Mamdani went on WNYC to defend hiring more NYPD officers, per Gothamist Daily, which also ran a strange and tender piece on the dead animals being found in Prospect Park. Casey Lewis at After School covered lunch shame and mall rats with field notes from the city's outdoor dining season. Consuming Collective's "Where to Eat in NYC this Week" landed alongside Emily Sundberg's Feed Me reporting on how a phone, WiFi, and a little creativity now constitutes a viable OnlyFans business in the city.

The combined picture is a Mamdani New York where the mayor is governing from the policy center, the food scene is back to being intensely local, and the side hustle economy is more openly transactional than at any point since the late nineties.

Ideas Worth Reading

Outside Interests

Data Worth Noting

Three Takeaways for You

The Anthropic IPO is not just a liquidity event. It is the moment the AI cycle becomes a public-markets story, which means the next year of "is AI working?" will be priced in basis points at Alphabet, Microsoft, and a soon-to-be-listed Anthropic, not in benchmark charts on X. Watch the cost of capital, not the model leaderboards.

The labor-impact debate has clearly broken into two camps that no longer share data. Edwards and Noah Smith are looking at the same economy that Anthropic, OpenAI, and Sacra are looking at and reaching almost opposite conclusions. The honest answer is that both can be true for another year, but the gap closes once the IPOs price.

If you only read three pieces, I would suggest Casey Newton's interview on the AI jobs-pocalypse for the labor argument that will frame the next cycle, Ben Thompson's Mythos, Muse, and the Opportunity Cost of Compute for the capital frame everyone in tech is about to converge on, and Bill Kristol on the July 4th hijack for the political stakes of the month ahead.