A short week in review
What I read this week
I missed my normal friday schedule because I was sick. Feeling better now but this week’s letter mostly links (what I read) and light on my own writing (I didn’t do any!). We’ll be back on our normal schedule next week.
While sick I binge-watched the adaptation of Three Body Problem which reminded 1) of how much I liked the first book and 2) how much I like the plot device of needing to think/strategize across decades or centuries bc of the constraints of interstellar travel (and warfare). It’s all plans within plans because you can’t easily pivot. 3BP, Hyperion, Foundation, Seveneves all do this well.
I Read
If GLP-1 Drugs Are Good For Everything, Should We All Be on Them? The short answer to this is “yes” and it seems like we all probably will be. Whether the weight loss drugs impact things like Alzheimers and Parkinson’s directly or because those conditions are ultimately downstream of being overweight is TBD. But I don’t think it will, with the benefit of hindsight, look like an overstatement to say that GLP-1s are the most important medical breakthrough of the century.
There's plenty of water for data centers. There is a brewing water-based backlash to AI which is a) wrong and b) largely rooted in the fact that most people don’t really know/understand/care to know what the embodied water usage of most activities is. Is AI putting a strain on the power grid? Yes. But framing it as a water issue misses the point both directly (it’s about power) and indirectly (if we had more power, we could use non-potable water). Everything always comes back to energy prices!
The Boring AI Questions That Actually Matter Those questions are: Open vs closed, why are some models better, does safety matter, and should we sell chips to China? Can is back! Margins has long been one of my fave newsletters. Now that he’s unemployed he has time to write again. I’m excited to get into this with him IRL in a couple weeks over dinner.
Confessions of a Laptop Farmer: How an American Helped North Korea’s Wild Remote Worker Scheme Much deeper dive on the question of what North Korean fake employees actually do and it seems to be more mundane than I first expected. It’s about doing work and getting paid vs doing sabotage or exfiltrating secrets, at least most of the time and for non-crypto companies.
Increasingly, the scheme includes a kind of industrial espionage, as the workers also quietly exfiltrate sensitive data. Crypto companies are especially rich targets. Sometimes the workers turn to extortion, threatening to reveal sensitive secrets if they aren’t paid a ransom. But security researchers have found that most of the workers raise money for the North Korean government simply by striving to do average work and collect their pay. Humdrum or not, the scheme requires a steadyish supply of corporate marks. Some employers, Barnhart says, are more scandalized than others to discover they are one. “I think one of the biggest things we run into with some customers is they say, ‘Who cares?’ ” he says. “We’ve seen instances where they were like, ‘Do we really have to fire them? Are you certain they’re North Korean? Because this guy’s good.’ ”
Reindustrialization, capital crowning, bougie gyms, and mediated selves. There’s a meme going around/a sentiment pervading that the only companies that can raise are 1) high flying AI companies, 2) companies that can convince they will be high flying AI companies 3) can’t miss founders 4) American dynamism/re-industrialization/whatever you want to call it. The meme is astutely observer and right (and probably self-reinforcing) but not good.
IMO this is the legible/illegible divide in action whereby companies are either extrinsically legible (building "the right" businesses) or intrinsically legible (just crushing, PMF, etc.). The challenge is early and in the "wrong" categories bridging to intrinsic legibility. But that's the point of risk capital!
I continue to be excited to bet where others won’t/can’t on the belief that I can fund companies to the point of either narrative legibility (pre-consensus to the right themes) or intrinsic legibility (they’re just crushing and can’t be ignored).
Tech for Abundance
New York is clearly at a crossroads and I see tremendous potential energy to harness for change. Next month I’m hosting an event with Abundance NY, Lux Capital, and Company Ventures in NYC. We’re bringing the startup and tech community together to share ideas and hear from startup founders and leaders in government. We’ll be announcing the agenda and speakers as we get closer to the event.
New York City is in flux, politically and technologically. Old paradigms of liberal governance are being challenged by adherents of abundance as AI disrupts industries.
The City should be a great place to build a business AND should take full advantage of our entrepreneurial spirit to make government work better.
This event has garnered a ton of interest, which is super exciting - we had to change the venue to account for more people! We’re expecting a group of ≈150 people vs the 50-75 we first imagined when we planned an August event.






HYPERION
one of the most underrated sci-fi sagas out there