June 17 – CEOs

Cats, Glasses, Glory, and the Menu

The feed moved in short, sharp bursts rather than sustained lines. Most of the day passed in near silence from the largest names in the cohort. Then, toward evening, a smaller number of voices stepped forward with reactions that felt less like original dispatches and more like precise replies to something already in the air.

Marc Andreessen threaded several of those replies together. He surfaced a clip of Nassim Taleb laying out the familiar pattern in which the first movers in a transformative technology rarely capture the lasting value. The early car makers, the early airlines, the early personal computer companies mostly vanished or were overtaken; Andreessen let the parallel sit without much added commentary. Later he posted his own unease at the revelation that sitting and former presidents had personally pressured Spotify to sideline Joe Rogan over vaccine questions that later checked out. He noted, almost in passing, the long-running policy paradox in which decades of trying to make college affordable had produced the opposite result while ordinary market competition had made high-quality televisions cheap. He shared research that turns the firing patterns of monkey visual neurons into human-language descriptions of the images that trigger them, and he reacted with plain surprise at the method. He posted variations of a single savage image—three overweight cats in a trench coat—captioned with the line that the figure in question could not fool him. And he closed one of the later posts with the simple observation that ground truth beats snap judgments when the task is understanding what anything actually looks like from the inside.

Chamath Palihapitiya moved between the macro and the immediate. He amplified a New York Times essay that pushed back against the recurring claim that artificial intelligence will end the world, adding only that it will not. He wrote a single, unadorned sentence rejecting remote work. He described, with evident satisfaction, the team he had assembled—its character, its path from internship to full contribution, its particular mix of talent—and left an open door for others who wanted to work inside that frame. He noted how quickly open-weight models were closing measurable gaps with their closed-source counterparts and predicted that recursive reinforcement learning would compress the remaining distance still further. He also pointed people toward live demonstrations of a new enterprise software layer presented as the emerging control plane for coordinated work.

David Sacks used the day to correct the record on an earlier exchange. He shared a New York Times opinion piece on the dangers attached to current frontier efforts at OpenAI and Anthropic, then returned with a longer post that pushed back against selective quoting of his own prior comments. The core claim he defended was straightforward: the cyber risk is real, every organization that maintains large codebases should treat it as such and move to harden systems now, and the defensive work requires cooperation that has been complicated by unnecessary alarmism and confrontational posturing from one of the companies involved.

Zvi Mowshowitz stayed inside the technical and policy layer. He laid out, with characteristic precision, how certain classes of jailbreak could be blunted by narrowing the scope of the classifiers that currently block them. He noted the leverage that export controls on advanced chips continue to represent. He observed that some of the largest stories in the space were receiving oddly little front-page attention despite their scale. And he described one high-valuation transaction as the latest instance of a familiar pattern: selling expensive equity to acquire exposure to the next scarce asset.

Brad Stulberg offered the day’s clearest counter-note to the prevailing tone of friction and risk. He posted Lionel Messi’s reflection after a World Cup hat-trick—Messi saying he recognized in Rafael Nadal the same drive to give everything and to enjoy the work while doing it. Stulberg tied the comment directly to the two ideas at the center of his recent book: that genuine enjoyment functions as a durable competitive advantage, and that caring enough to try hard remains both legitimate and effective.

Emmett Shear wrote in a quieter register about practice and form. He described the value of rehearsing arguments that run from arbitrary premises to arbitrary conclusions, including the discipline of arguing the opposite case with equal rigor. He suggested that anchoring in one resonant tradition while still sampling from others can be a workable path. He acknowledged both the immediate fit of syncretic approaches and the hazard that important elements get discarded before their necessity becomes obvious. He noted that certain sources appeared roughly equal in the precision of their language and that words, treated as shapes, already function as concepts even when the reverse is not always true.

Alex Wissner-Gross’s contributions were mostly minimal, but two lines stood out. He posted the old, brutal observation that if you are not at the table you are on the menu. He also wrote, without elaboration, that he was on the road again, making strangers rich.

What accumulated across these posts was not a single argument or a shared mood but a set of precise, localized interventions.