New York, New York
— March 9, 2026

If AI Is a Bubble, It Won’t Be Because People Were “Stupid”

It’ll be because the ROI is unknowable right now.

We recently covered a conversation between Andrew Ross Sorkin and former Goldman CEO Lloyd Blankfein that hit three of the biggest forces driving markets right now: AI CapEx, geopolitics, and private credit.

Here’s the clean takeaway: we’re living in a moment where the future is being financed before it’s fully measurable. That’s exciting. It’s also inherently unstable.

---

1) AI spending boom: “Not evil… just unknowable”

After managing through the housing bubble and financial crisis, Blankfein’s framing comes down to this: If AI ends up being a bubble, it won’t be because executives were malicious, or the market was morally compromised. It will be because they placed a massive bet on a return that can’t be proven in advance.

That actually matches what we’re seeing in real time: hyperscalers are building AI infrastructure at a scale that feels historic, and the payback period is still being written. A lot of this investment will work. Some of it won’t. And the market will probably only know which was which after the cycle turns.

That’s not “stupidity.” That’s the nature of frontier CapEx.

---

2) Iran conflict: measured, but statistically optimistic

On Iran, Blankfein was serious but not alarmist. The core idea was that crises of this magnitude tend to become self-limiting, because the downside consequences are too severe to sustain for long.

One detail he flagged as worth watching is the way previously non-allied countries can start coordinating when a shared threat emerges, including around air defense. He framed that as a sign that escalation pressures can also create coordination pressures, which in turn can shorten the duration of the worst outcomes.

“Maybe that’s optimistic,” was the vibe, “but historically optimism has a statistical basis.”

---

3) Private credit: concerning, but not 2008 in structure

Blankfein also spent time on private credit, and this is where the nuance matters.

He’s concerned about the risks building in private markets and the potential for a “reckoning” type moment. But he drew a sharp distinction from 2008: that was a banking crisis where the system seized up and only government intervention could restart credit creation.

Private credit is different. It’s not the same “lender of last resort” structure. It’s not the same plumbing. The risks are more about opacity, liquidity, valuation marks, and who is holding what than a classic bank-run dynamic.

In other words: it can still be painful, but unlikely a another 2008 scenario.

---

Bottom line

This conversation landed because it hits on everything we see in markets today with deep expertise of the past.

• AI may be transformational, and it may still be overbuilt in pockets.
• Geopolitical shocks can be severe, and they can still be self-limiting.
• Private credit can be a real stress point, and it doesn’t need to be “2008 again” to hurt.

If you’re feeling the market get harder to predict, you’re not imagining it. We’re watching three big forces overlap at once.

Be good.