- Who’s paying AI bills? https://marketbites.substack.com/p/ai-is-eating-the-world-but-whos-paying
- When people say “AI feels like a bubble,” they’re not doubting the technology but the math behind current valuations.
- The average AI stock now costs about 22 times its yearly sales, up from 7 times three years ago. In plain terms, investors are paying over three times more for every dollar of revenue than they used to.
- Big cloud companies trade at about 35 times their profits, chipmakers around 45 times, and software firms selling “AI copilots” are priced at 70 to 100 times what they might earn in the future.
- Public AI companies are worth over $15 trillion, but they make only about $170 billion in actual profit. That’s a price-to-earnings ratio of roughly 90 times, far above what’s normal. For comparison, during the 2007 housing bubble, homebuilders traded at only around 25 times earnings.
- So yes, the AI market might be overheated, but it’s not built on hype alone. There are real products, users, and massive infrastructure behind it.
- If the bubble bursts, AI won’t die. The prices will just come back to earth, and the shock will spread through the economy like other tech booms before it.
- Company as a machine for doing stuff https://www.notboring.co/p/the-company-as-a-machine-for-doing
- At their best, companies are tools for doing the work a founder actually cares about, but at a larger scale and with people who want to chase the same mission.
- Many startups drift into being vehicles for status or wealth, used to collect the resources people think they need before doing the thing they really want to do.
- The more useful path looks like what Elon did after Zip2 and PayPal: he poured his winnings into Tesla and SpaceX, not to get richer but to make the work possible. You can’t build electric cars or rockets alone in a garage. The money mattered because it let him gather people who were motivated by the same work (also Jeff Bezos).
- What you’re building doesn’t have to be a lifelong dream, but it has to be big and open-ended enough to stay interesting for decades, because that’s how long important things take.
- People are less productive in higher temperatures https://unchartedterritories.tomaspueyo.com/p/mountains
- People are less productive in higher temperatures.
- Each +1°C reduces GDP per capita by about 8.5%.
- Warmer temperatures lower both agricultural and industrial output.
- Every +1°C above 27°C cuts manual labor productivity by around 4%.
- Each +1°C increases absenteeism by roughly 5%.
- A +2°C rise leads to about a 10% drop in memory, reaction time, and executive functioning.
- People sleep worse in warmer conditions, further reducing productivity.
- Lee Kuan Yew likely understood this well, which is why he considered air conditioning a key factor in Singapore’s success. https://www.vox.com/2015/3/23/8278085/singapore-lee-kuan-yew-air-conditioning
- Pros and cons of Europe and US in everyday life https://x.com/patrickc/status/1908960227921424599
- Europe is better for everyday life thanks to dense walkable cities with real bike lanes, late-night cafes, beautiful old buildings, pedestrian streets, good bread, cheese, wine, church bells, pleasant pharmacies, proper bathroom doors, great trains, clearer motorways, better everyday regulations, and tastier cured foods.
- The US is better for comfort and convenience, with reliable air conditioning, stronger coffee culture, fewer cookie banners, faster internet, easier access to capital, open Sunday shopping, better showers, cleaner urban air, quicker government processes, more flexible labor markets, widespread small airports and pilots, strong hospitals (?), and craft beer.
- Building a search index is very hard because the web is huge, constantly changing, full of duplicates, and you have limited bandwidth to crawl and refresh it, so you’re forced into endless tradeoffs about what to fetch, how often to fetch it, and how to turn that messy snapshot into something searchable at scale. https://x.com/AravSrinivas/status/1971626155594928313/photo/1
What I’ve been reading (W46, 2025)
AI markets, why build companies, effect of temperature on productivity, daily life in Europe vs US, and complexity of building a web index