2026W07
- Hedonism and Entrepreneurship in Barcelona — A potential acquisition dies because a vegan exec unknowingly eats mayo on patatas bravas. A critical migration stalls at 69% while the engineer responsible plays ping-pong. Having lived the Barcelona startup scene myself, this one hit close. The absurdity is the point — startup success is so much more arbitrary than anyone wants to admit, and the line between “almost rich” and “back to Monday” is thinner than your Series A pitch deck suggests.
- OpenClaw, OpenAI and the future — After 13 years building PSPDFKit, Steinberger explicitly chose not to build another company. Instead he’s joining OpenAI and turning OpenClaw into a foundation. I keep thinking about the tension here: he wants to keep things open and independent, but the path to maximum impact runs through one of the most powerful closed labs in the world. Whether that trade-off holds up depends on how seriously OpenAI takes the foundation model. Time will tell.
- They’re putting blue food coloring in everything — It’s not about blue food, obviously. It’s about how unwanted things get normalized — first one restaurant does it, then “all the best restaurants” do it, then your friend sneaks it into homemade food and tells you you’re overreacting. The detail that got me is the protagonist saying “I just think it tastes weird” and being told “most people say it’s just fine.” I’ve had that exact conversation about too many things in tech.
- Why I joined OpenAI — The turning point wasn’t a demo or a benchmark — it was his hairstylist casually mentioning she uses ChatGPT all the time. She recognized ChatGPT as a brand more readily than Intel, where Gregg was a Fellow. That contrast says more about where computing impact has shifted than any industry report. The environmental framing is interesting too: at this scale, performance engineering isn’t just cost optimization, it’s resource consumption with real planetary consequences.
- GitHub Actions Is Slowly Killing Your Engineering Team — The fact that an entire cottage industry of startups exists solely to fix GitHub Actions’ slow runners tells you everything. The comparison to Internet Explorer is spot on — it wins because it ships with the thing, not because it’s good. I’ve felt the pain of debugging through four pages of loading spinners to find a failed step, and the escape hatch of “just write a bash script” that inevitably becomes 800 lines of unmaintainable CI logic. Default integration is a powerful moat, even for mediocre products.
- 2025 letter — Wang’s framing of the US economy as “a highly leveraged bet on deep learning” stuck with me. The Xiaomi vs Apple comparison is brutal: Xiaomi shipped an EV in 4 years while Apple spent 10 years and $10B before giving up entirely. His concept of Silicon Valley’s “soft Leninism” — groupthink disguised as meritocracy — is uncomfortable because it’s hard to argue against when you see how the industry moves in lockstep. Not sure I fully buy the symmetry he draws between SV and the CCP’s self-seriousness, but it made me think.