New newsletter: THE FOUR HORSEMEN OF THE AI BUBBLE APOCALYPSE I am neither anti-AI nor certain that AI is a bubble. But the last four weeks have made clear that the AI buildout now faces a very clear quadruple-headed risk hydra. 1. A spending risk, as the hyperscalers run low on cash and take on $170b in annual debt—which is more than the projected UK deficit. 2. A revenue risk, as open-weight models threaten to compress the margins of frontier labs ... and as AI become the sort of internationally competitive asset-heavy industry requiring stable and determined long-term policy consistency, which is arguably China's competitive advantage. 3. A political risk, as anti-AI populism becomes one of the easiest applause lines, even as AI becomes a more and more foundational pillar of US economic growth 4. A technological risk, as the frontier labs bear down on RSI, which I think could significantly change the basic business model of the labs, as compute costs rise and rise for a set of super-advanced models that are fit for, and affordable to, a small minority of users (in, eg, cyber security) In one sentence: The capabilities of AI are becoming more powerful, while some economic underpinnings of the AI buildout—and, as we’ll discuss, the political support for AI—are becoming more vulnerable. Today's (long, 5k word) piece deeply considers each risk and also—because over-confidence in this space is typically a sign that you're not thinking hard enough—I offer the strongest reason to think each risk might be overblown.

I knew she was a diligent scholar and talented writer, but didn’t realize Emily Wilson was such a good interview. If you read her version of the book, this would be a good listen before heading out to see the movie.

New pod: EMILY WILSON ON CHRISTOPHER NOLAN’S ‘ODYSSEY’ I talked to the famed “Odyssey” translator Emily Wilson about why she thinks The Odyssey missed a mark and why Nolan’s Odysseus is almost unrecognizable from Homer’s. To borrow from Damon’s filmography, Homer’s protagonist is more heroic Tom Ripley in armor—a master of deception, disguise, and twisty morals—while Nolan’s is more guilt-laden PTSD Jason Bourne. She also acknowledged that the moral landscape of Homer’s Odyssey is so foreign and nearly alien to us that it’s hard to imagine a summer blockbuster that that didn’t make some changes. We also talked about the art of translating ancient texts, how the definition of what makes for a “good translation/adaptation” has changed in the last few centuries, and why we care more about historicity and verisimilitude than we used to.

nolan as the male taylor swift is funny - genius artist who guards the last dying embers of the monoculture - a popular mainstream hit maker whose work is also critically acclaimed and wins big awards - ludicrously productive and famously known to take care and pleasure in actually writing their own work and they're opposites in ways that are kind of gendered, too, in that the thing that makes them special is distinct. taylor swift's best music is memoiristic and about the emotional inner life of non-married women whereas christopher nolan's stuff is almost entirely unromantic, my favorite protagonists of his are dads, the characters don't really talk about their feelings unless they're telling you the theme of the movie (oppenheimer: let's have sex while i foreshadow the trinity test with this shiva quote in spontaneously translated sanskrit), and he excels at macro-level narrative spectacle rather than emotional interpersonal dialogue

this is maybe a shade too provocative a framing, but ... the wild thing about folks in and close to govt talking about banning foreign tech and taking direct shares in frontier labs to protect them from short-term profit squeezes on the way to global domination is that you can almost frame a question being posed here as something like: "can the US government become China faster than Kimi can become Claude?"

This is a fascinating, honestly kind of shocking post. Of course, lots of VCs are quote tweeting it. Look at what it says! - "American companies need a legal way to turn American frontier capability into cheaper, ownable models." - "Frontier labs could sell structured training rights to qualifying Western and allied companies" - "If a voluntary market does not develop, access could eventually become a condition attached to major federal AI contracts, for example." So...Frontier Labs should hand their capabilities to other American companies at low cost. If they won't, the government should force them. A transfer of value from the companies building the frontier to the companies who don't want to pay. Framed as a national security requirement. The VC drumbeat is getting louder. Very interesting times.

This is well done. But I want to add something: Age! According to the DSA's own surveys, the share of members who are Millennials rose from 7% in 2013 to 60% (!?) in 2021. If you include old Gen Z and young Gen X, I'll bet ~80% of DSA is between 18 and 45. DSA is extremely Millennialish. Many things happening here, but one thing is that the Great Recession made self-described socialists out of a lot of Millennials. This cohort had no political power in 2009. But now they're in their 30s and 40s and a much bigger part of the electorate, especially in cities, where they're pissed off about housing, especially if they aren't rich. Another way of putting this is that it's useful to think about American socialism as a cohort effect.

On a Hollywood economic vibe shift: The first two decades of the 20th century were dominated by what you could sort of simplify as the Iger hit formula. (1) Buy familiar IP; (2) Establish franchise; (3) Extend franchise; (4) Repeat. This was so reliable that for years, practically every film in the top 10 had a sequel-number in the title. Iger was brilliant, but the Iger formula is faltering. Remakes and sequels of Moana, Star Wars, Minions, and Supergirl have all disappointed recently. But every Nolan movie these days is a billion-dollar event, Sinners was a monster, and Villeneue makes almost all bangers. That this is coinciding with the Obsession/Backrooms phenomenon is interesting, too. @MattBelloni: "I'd rather have the next commercial movie from the five biggest directors (Nolan, Cameron, Coogler, Gerwig, Villeneuve) than the next installment of almost any piece of well-worn I.P. Wouldn’t you?" The optimistic gloss on this would be that I, a person who never read comic books and don't like Marvel very much, am much more interested in seeing movies that were greenlit on behalf of the ingenuity of the director or the script, rather than the mathematical inevitability that "we did Ant-Man 2, so we're doing Ant-Man 3." It's also cool that this shift is happening during a relatively buoyant year for ticket sales. The risk for Hollywood, I think, is that the bar for success is getting higher; or, put differently, I think maybe the floor is lower. If you owned the Marvel or Minions franchise in the 2010s, you had a relatively high floor for the next Marvel or Minions movie, bc you knew a certain number of people would turn up even if the Rotten Tomatoes number was 60. But what happens when you're sort of starting from scratch every time? If you're a studio, it's really hard to do 10 Obsessions in 10 straight years! That was a weird fucking movie. Bottom line, the Iger-Nolan vibe shift in Hollywood is great for people who like original big swings and blockbusters that also try to win Academy Awards. But, despite an awesome 2026 for the Return of Movie Theaters narrative, it's sobering to remember that sui generis hits are rare, Nolan is rare. If the bar for hits is "it better be genius," that's an interesting business problem, bc genius is rare ... maybe even rarer than familiar IP.

this is such a funny way of putting it. and it's so true of the models. it's rare that a product's quality oscillates in such a way that the underlying financial/capex health of the company is visible to the consumer. like if you could guess chipotle's YTD stock movement by just watching how much chicken they put in your burrito

again, big picture, donald trump won an election about affordability and then he - passed tariffs (which raise consumer prices) - expanded the deficit w a ginormous corporate tax cut (raises interest rates) - waged a stupid war (raises gas prices) - and now refuses to sign a bipartisan bill to reduce housing costs bc he's protesting senate republicans for failing to pass voting restrictions ... while making $2 billion from crypto rug pulls that transferred income from his supporters to his bank account americans hate the president for very good reasons

A theory I'm testing out about AI, work, and the future of companies ... I have for months been obsessed with Alfred Chandler's The Visible Hand, a classic business-history book that argues that 19th century technology—the telegraph, train—created an explosion of real-time sales and inventory-information challenges, which required the creation of a new org charts with something called "middle managers." Thus, technology birthed managerial capitalism. Thematically, Chandler's insight is that technology doesn't just change jobs; it changes how jobs are organized in relation to each other; which is to say, tech changes how companies are structured. So, how might AI restructure the firm? Here's one stab at an answer: The telegraph/train created a profusion of information, which required managers. But AI's greatest skill is synthesizing information that's already been created. I think that's one reason why new companies are getting rich faster with unusually small staff. The "visible hand" of professional managers is being replaced, at the margin, by the "invisible hands" of digital platforms and AI assistants that don't so much destroy work as they allow new firms to build lean human-AI workflows that are qualitatively different than the larger bureaucracies in their industries.

There are some center-to-right wing accounts I follow that are more offended by the possibility that some public health sources are over-estimating the death count from Musk carelessly destroying US global health programs than by the core fact that Musk really did carelessly destroy US global health (and bragged about it relentlessly) in a way that clearly killed people. These people know who they are, and they're wrong.

At this point, it’s better to assume that all longform tweet essays are written by AI (while reserving the right to be surprised if it’s a exuberantly human and spiky-prosed piece) rather than assume they’re all written by people and being constantly shocked by the triple-digit pangram score