Episode summary
278: Paul Kedrosky, Managing Partner, SK Ventures
Summary
Paul Kedrosky argues that the AI capex boom is unusual because it combines four drivers at once: a powerful technology story, a real estate-style data center buildout, loose credit, and unusually strong government support tied to U.S.-China competition. He says that combination makes the moment look like a “rational bubble” in which each participant can view their own behavior as sensible, even as the system as a whole becomes fragile.
He compares the current cycle with the dot-com bubble, the Global Financial Crisis, and Japan-style balance sheet deleveraging. In his view, this is not just an equity story or a credit story; it is increasingly a private-market phenomenon, with huge valuations in companies like SpaceX, Anthropic, and OpenAI and a growing risk of prolonged deleveraging if the cycle turns.
Kedrosky also focuses on concentration and contagion. He says venture-backed companies are the largest spenders on AI on a per-employee basis, and that the earliest users of large language models were unusually concentrated in software, making it dangerous to extrapolate to broader white-collar labor. He thinks Nvidia functions like the “central bank of AI,” but warns that the system is moving into tightening mode as credit to neoclouds and other counterparties becomes more constrained.
Key takeaways
- Paul Kedrosky says the AI boom is unique because technology, real estate, credit, and government policy are all reinforcing the same buildout at once.
- He argues that private-market valuations in SpaceX, Anthropic, and OpenAI are already comparable in nominal terms to the peak NASDAQ valuation at the dot-com top.
- He says venture-backed companies are the biggest AI spenders per employee, which makes the current demand base unusually concentrated and fragile.
- Kedrosky thinks Nvidia is acting like the “central bank of AI,” but he says the system is now entering a tightening cycle for neocloud financing.
- He warns that the bigger tail risk may be a repricing of white-collar labor itself, not just a reset in GPU prices or token demand.
Notable quotes
“it combines each of those components, each of those pillars, not quite an equal way, but in a very comparable way”
“There's a real estate component to many people participating in the AI moment”
Topics
- ai capex cycle
- ai investment booms
- dot-com bubble parallels
- corporate deleveraging
- credit supply and ai financing
Companies and tickers mentioned
- NVDA
- SK Ventures
- MIT
- Nvidia
Watch the episode
Watch “278: Paul Kedrosky, Managing Partner, SK Ventures” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Alpha Exchange
-
Consumer Stress Hiding Beneath a Strong Market -
Financing the AI Boom: Why Power Assets Appeal to This Credit Investor -
What Does a MOVE of 100 Actually Mean? -
Why Markets Keep Creating New Opportunities for Alpha -
When Your Prime Broker Fails: Lessons from Lehman -
How the AI Boom Is Distorting Corporate Debt & Bond Markets
Alpha Exchange, this episode and its artwork belong to Dean Curnutt. Octagon is not affiliated with the show; this page is an independent summary.