Four For Friday | June 26, 2026
LF227 | Europe's future, judgement infrastructure, universities as monasteries and the 13% impact premium + video of the week
Welcome to this week’s Four For Friday. Four topical stories on systems change, healthspan and AI for impact. Enjoy!
1. Europe’s future, written in Washington
A sobering scenario traces Europe’s slide into AI irrelevance through 2031. Told via two fictional characters, a Brussels official and an émigré founder, it grounds the fiction in hard numbers. Europe hosts just 5% of global AI compute; America hosts 80%.
The largest US supercomputer runs at 1,250MW, Europe’s at 83MW, a 12.4-times gap that barely narrows by 2031. Von der Leyen’s headline €200bn InvestAI fund is mostly repackaged money, against $400bn in US hyperscaler spending in 2025 alone. The EU’s goal of attracting €200bn by 2036 amounts to a quarter of America’s quarterly capex.
Sovereignty mandates backfire: “buy European” agencies end up paying ransoms when cheaper, weaker models cannot defend them. The compounding lesson is that leverage, the ASML chokepoint aside, evaporates when you arrive late.
The 2034 epilogue, in which a regretful Caroline is interviewed by an AI, argues the failure was structural, not villainous: institutions responding rationally to incentives unfit for the moment.
The So What: Latecomers don’t get sovereignty. Europe needed to build compute like a country at war.
2. Judgement infrastructure
The bottleneck in AI adoption is no longer technology but judgement.
This HBR piece argues that most firms share the same models and infrastructure; what separates leaders from laggards is “judgement infrastructure”, codifying the tacit reasoning of expert staff so agents apply it consistently. The cases are pointed. At AWP Safety, a one-person compliance team (within a 9,100-strong firm) built agents that save hundreds of hours a year. A finance controller using Claude now covers work that once needed ten. Ramp, used by 30,000 companies, trains every hire to build their own tools. Expertise, once trapped in senior heads, becomes portable.
The So What: The next moat isn’t the model. It’s writing down how you actually think.
3. Universities - don’t become monasteries
Geoff Mulgan warns universities against complacency, invoking the dissolution of the monasteries: institutions that looked permanent, then weren’t.
The sector has never been richer. Student numbers have swelled from 13 million in 1960 to over 260 million, with global turnover nearing a trillion dollars and more than 50,000 universities, up from perhaps a thousand in 1900. Yet scale breeds fragility. AI threatens to upend formal instruction and research funding; political attacks mount; and the UK earnings premium is slipping.
Mulgan’s prescription: a new social contract built on challenge-based work, place-shaping, lifelong learning, and metacognition, rather than complaining about insufficient love.
The So What: Richer than ever, yet so were the monasteries before Henry VIII came calling.
4. How to price externalities? 13% apparently
The impact investing world finally has a hard number for the "tradeoff" between impact and return. A landmark study by the Miller Center for Global Impact found that the cost of an impact-first approach stands at exactly $0.13 per dollar deployed. This "premium" isn't waste; it's the cost of delivering impact where markets traditionally fail.
The So What: This approach suggests deep social impact costs exactly $0.13 more per dollar deployed. Worth it?
Video of the week: Michael Truell, Cursor
This video has been doing the rounds recently and is worth a watch. This is a company that had its first launch beta in 2023 and just sold for $60bn. Cursor’s 25-year-old CEO and founder is incredibly articulate, thoughtful and humble. And more substantively, he makes a persuasive case for not worrying about entering a space that already feels crowded if you make a high-quality product.
That’s all for now, happy weekend, everyone.
- Stephen