ZD 26.28: The Great Moat Hunt
AI labs are working very hard at not becoming commodities.
📈 Escape From the AI Commodity Trap (Arvind Narayanan)
Selling AI models is a bad business. They can be described as undifferentiated products with high capital cost and little leverage over customers. They need to move up the value stack. The labs need to shift from models into products, intelligence and even digital workers to have a chance to recoup the billions invested.
🪙 Thinking About Token Pricing (Benedict Evans)
Tokens are supply constrained today, but supply crunches rarely last. In recent technology waves like telecom and software, value accrued closer to the top of the stack, not in the infrastructure layer, despite billions of investment. It is another sign that models are not the moat some people want them to be.
🧪 The Chemistry of Chips (ChinaTalk)
Making chips seems calm and sterile, but, at the atomic level, it is an extremely violent process. You need a collection of rare, highly reactive and deadly chemicals to make it work. Like everything else, China controls the supply of these critical gasses.
🧑💼 Conway’s Law Applies To Agents, Too (Joe Reis)
Organizations design things that match their communications structures, and agents are no different. You still need to figure out product and data ownership to succeed.
⚓ Winning Below the Waterline (First Breakfast)
The USS Constitution outgunned every frigate afloat thanks to a system of diagonal timbers making it capable of handling that many guns. The same story repeats through 100-octane fuel and air-to-air refueling capabilities today.
🚑 Why American Ambulance Rides Are So Expensive (David Oks)
Most ambulance operators lose money. A 1965 Medicare decision to reimburse carriers per ride effectively put operators into the business of selling the equivalent of an option on healthcare paid for by insured customers like you.
💰 Trump Accounts and Related Ideas (Damped Spring)
Like everything in finance, it pays to read past the label. The debt-funded equity purchase will do nothing but bid up stock prices as a tiny flood of new money hits the market.
🕰️ Medicine is Moving from Calendars to Clocks (Eric Topol)
Medicine is learning a lot about how we age. It is not a linear process as one might imagine, nor is it the same in everyone. AI is letting researchers sequence proteins in ways that were not imagined even recently. Treatment protocols are being developed to catch organ degradation early before it becomes critical.
😵 The Decline of Deviance Revisited (Adam Mastroianni)
Teen smoking, drinking, crime, and pregnancy have collapsed since the 1990s. Is prosperity to blame?
📖 How Malcolm X Taught Himself (Sam Rinko)
Malcolm Little walked into prison barely able to write a letter and came out one of the century’s great orators. His self-directed studies while incarcerated surface guidance such as study with a cause and read wide before reading deep.
⚽ American Soccer’s Self-Inflicted Ceiling (Jay Shemenski)
America doesn’t play soccer like a nation of 330 million because the American way of soccer limits the talent pool to the much smaller number of Americans who can afford travel teams and private clubs.
The Look
The history of macOS, told through the icons you used.
Did you enjoy reading this post? Hit the ♥ button above or below because it helps more people discover great Substacks like this one and it helps train your algorithm to get you more posts you like. Please share here or in your networks to help us grow!