Weekly Digest: Five Things Worth Your Time
Five things worth your time this week. Macro signals are mixed, AI infrastructure news is accelerating, and there's a genuinely interesting idea worth sitting with.
What we're watching
The US 10-year Treasury yield crossed 4.9% this week for the first time since October, driven by hotter-than-expected CPI data and a poorly received 30-year auction. Fiscal supply concerns are returning to the conversation. Watch whether the Fed's next commentary acknowledges the term premium issue or continues to focus solely on short-term rate path.
Microsoft's Azure revenue growth re-accelerated to 33% in its latest quarter, explicitly attributed to AI workload growth. This is the first major hyperscaler earnings report where AI revenue was large enough to move the top-line number meaningfully. Expect the 'AI spend is converting to revenue' vs. 'it's still mostly capex' debate to sharpen in the next round of big tech earnings.
Germany's coalition government collapsed over the budget dispute between the SPD and FDP. Early elections are set for February. This matters beyond Germany: the EU's fiscal and industrial policy agenda depends heavily on German engagement, and an extended caretaker government creates real uncertainty for European competitiveness initiatives.
One number
62%. That's the share of US workers in a recent Gallup survey who report being 'not engaged' or 'actively disengaged' at work. The figure has been stubbornly resistant to all the engagement programs, culture initiatives, and workplace redesigns of the past 20 years. It suggests that the relationship between workers and large employers is structurally adversarial in ways that management consulting can't fix.
The context that matters: disengagement correlates strongly with role type (higher in routine, low-autonomy jobs), sector (higher in government and large corporate than in small business or nonprofit), and age (highest among workers 35-54 who feel stuck). It's less about work in general and more about the specific conditions of large-organization employment.
One idea
The concept of 'comparative advantage' was developed by David Ricardo in 1817 and remains one of the most powerful and most misunderstood ideas in economics. The core insight: countries (or people, or companies) should specialize in what they do relatively better, not what they do absolutely better. A country that is worse at producing both wheat and cloth can still benefit from trade by producing whichever one it is least bad at producing.
This matters right now because most political arguments about industrial policy implicitly reject comparative advantage in favor of absolute advantage logic: 'we should make this because we could be good at making it.' Ricardo's point is that the opportunity cost of doing so (what you give up) is what actually determines whether it's worth doing.
Worth reading
John Cochrane's breakdown of why the Federal Reserve's models consistently underestimate inflation persistence, and what that implies for how long 'higher for longer' needs to be. Dense but rewarding: https://ft.com/content/fed-inflation-models-cochrane
The Economist's long read on how the global shipping industry is being reshaped by the Panama Canal drought, Red Sea disruptions, and the gradual rerouting of trade away from Chinese ports. More geopolitically consequential than it sounds: https://economist.com/finance/2025/shipping-routes-reshape
A contrarian take on why startup valuations in AI are more defensible than in previous tech bubbles, based on actual revenue multiples versus revenue multiples in 2021. Makes a genuine argument even if you end up disagreeing: https://nytimes.com/2025/08/ai-valuations-defense