Practicing Wisdom — Issue #15

A weekly distillation of the most interesting things I explored, learned, and thought about.

1. What I Learned This Week

Theme: Cycles vs. structural shifts.

Eric Cinnamond’s “Cycles Anonymous” is a reminder that markets still move in rhythms — booms breed complacency, leverage hides fragility, and late-cycle confidence feels indistinguishable from structural change. The temptation isn’t ignorance; it’s rationalization. “Maybe just a taste.” That line hit. Cycles don’t disappear — they just seduce you into thinking they have.

On the other side, Dror Poleg’s “Welcome to the Jobless Boom” argues something deeper: what if this isn’t just a late cycle — what if the production function itself has changed? GDP rising while employment flatlines isn’t a lag — it’s a decoupling. OpenAI generating billions with hundreds of employees isn’t a blip; it’s a structural shift.

And then Ben Thompson’s “Thin Is In” adds the technological substrate: AI recentralizes compute. We’re returning to a thin-client world — interfaces local, intelligence elsewhere. Memory is scarce. Capital crowds into data centers. The edge device matters less; the core matters more.

Put together, a pattern emerges:

• Financial markets may be cyclically overheated.

• Economic structures may be permanently shifting.

• Technological architecture is re-centralizing power and profit.

Cycles are emotional while structural shifts are architectural. Are we confusing one for the other?

Cinnamond warns against believing “this time is different.” Poleg suggests it might be — but not in the way bulls think. It’s not that asset prices can defy gravity forever; it’s that output can increasingly detach from labor. Meanwhile, AI’s thin-client model suggests value will accrue disproportionately to those who own compute, memory, and the “Fifth Avenue” positions in data infrastructure.

Now we arrive at Josh Kushner’s interview. He talks about buying “Fifth Avenue” — category-defining assets — and holding them for decades. Markets may cycle, but compounding quality over long periods is the antidote to timing anxiety. If compute centralizes and productivity decouples from labor, then owning the platforms that mediate that shift is the Fifth Avenue strategy.

Markets cycle -> Technology restructures -> Capital concentrates.

The investor’s challenge is to stay sober in the cycle while being bold in the structural shift.

That’s harder than it sounds. Because late-cycle euphoria and genuine paradigm change can look identical in real time. As an aside, I also finished reading Andrew Ross Sorkin’s 1929 this week which made all of the above feel ‘same as ever’ in a lot of ways.

Sources Referenced

Cycles Anonymous — Palm Valley Capital (link)

Building Thrive Capital — Invest Like the Best (link)

Welcome to the Jobless Boom — Dror Poleg (link)

Thin Is In — Stratechery (link)

2. Key Distillations

• Cycles attack your discipline; structural shifts attack your models.

• When output detaches from labor, politics catches up before economics does.

• Thin clients create thick margins.

• The most valuable assets feel expensive every year — until they don’t.

• If everyone agrees risk is gone, it’s probably just hiding somewhere else.

3. One Contrarian Viewpoint

AI won’t kill jobs — it will hollow out status.

The popular narrative is labor displacement but the more destabilizing shift might be prestige displacement.

In a jobless boom, GDP grows, markets rise, but the social contract frays because fewer people feel essential to production. When fewer people are required to create more output, the question isn’t “Are there jobs?” but “Who matters?”

That tension shows up politically long before it shows up economically. The real disruption of AI may not be unemployment — it may be the psychological shift from being needed to being optional.

4. One Investable Idea

Memory is the new oil.

Thompson’s “memory crowd-out” thesis implies something underappreciated: AI’s bottleneck isn’t just compute — it’s memory bandwidth and infrastructure.

If AI workflows centralize intelligence in the cloud and require massive high-bandwidth memory, then:

• Infrastructure providers gain pricing power.

• Enterprises that own proprietary data moats become more valuable.

• Capital expenditure cycles shift toward compute-heavy infrastructure.

The Fifth Avenue of this cycle may not be applications — it may be memory, data center capacity, and proprietary datasets.

5. From the Archives: A Recall Highlight

“Brand is just memory plus trust.”

In a world where AI intermediates more decisions, trusted brands compound even faster. Memory — in silicon and in minds — remains the scarce asset.