The AI Trade Is Showing Cracks — And Three Podcasts Explained Why From Completely Different Angles

Aug 12, 2026 · Synthesized from 5 episodes across 4 shows

---

A 24-year-old's $45 billion AI fund collapsed to $10 billion in days. Meanwhile, a veteran venture investor is quietly betting that the *next* $100 billion AI company hasn't been founded yet. Both things are true at the same time — and that tension tells you everything about where we are in this cycle.

---

The Leverage Bomb Hidden Inside the AI Trade

Start with the most dramatic story of the week. The Prof G Pod dissected how Leopold Aschenbrenner's AI fund — running 5x leverage on momentum stocks — turned a routine market pullback into a forced fire sale to Citadel. Forty-five billion dollars became ten billion in a matter of days.

Galloway's framing is worth sitting with: leverage doesn't amplify skill. It eliminates the margin for being wrong. And right now, the structural conditions for more of this are quietly assembling. US leveraged ETF assets jumped from $120 billion to $200 billion between April and August — a 70% surge — with single-stock ETFs now comprising over half of all listings. Galloway notes this mirrors the exact structure that wiped out 70% of South Korean retail investors.

There's also a subtler accounting distortion running underneath all of this. Companies like Nvidia book AI revenues immediately, while the companies spending on AI infrastructure capitalize and depreciate those costs over five to ten years. The S&P's profit picture looks cleaner than it actually is. As short-seller Jim Chanos warns, financial fraud follows boom cycles with a lag — nobody investigates at all-time highs, only after losses mount.

Why the Smart Money Sees Something Different

Here's where it gets interesting. The same week Galloway was cataloguing leverage disasters, Benchmark partner Eric Vishria was on Invest Like the Best making the case that the dominant mistake investors make in big technology cycles is zero-sum thinking.

His historical lens: analysts in 2007 assumed AWS would consume all enterprise value. Instead, Azure, GCP, Snowflake, Databricks, Datadog, and Cloudflare each became $100 billion businesses. The market didn't get divided — it got larger. Vishria's bet is that AI follows the same pattern, and the investors panicking about which lab "wins" are asking the wrong question entirely.

The more specific insight — and the one that's easy to miss — is about where the value actually accumulates. Running large-scale AI models on identical open-source code and identical Nvidia hardware produces a 5x performance gap between specialized providers and standard cloud infrastructure. That gap comes from deep systems expertise, not hardware. Vishria argues that dismissing inference providers as commodity resellers is the same mistake analysts made about AWS. The hard part isn't the hardware. It's the execution.

The Structural Threat Nobody Is Pricing Into SaaS

Vishria's most provocative claim isn't about infrastructure — it's about what AI does to enterprise software moats. Database stickiness historically came from developers building against proprietary interfaces, making migration prohibitively expensive. AI agents now handle well-specified interface translation autonomously, turning a multi-year migration risk into a tractable, fundable project.

His conclusion is blunt: CEOs still executing pre-AI plans are destroying equity value daily, even while hitting their targets. The targets are fine. The moat is evaporating.

This is the tension that makes this week's conversation more interesting than a simple optimists-versus-skeptics frame. Galloway is warning about leverage and accounting distortions in the current AI trade. Vishria is warning about something slower and more structural — the companies that look safe right now because their switching costs feel durable.

The Behavioral Layer: Why Most Investors Will Miss Both Signals

Even if you've correctly identified the risks Galloway is flagging and the opportunity Vishria is describing, there's a third problem: your own brain. Investing for Beginners this week ran through the cognitive biases that cause investors to act on exactly the wrong signals at exactly the wrong moments — salience bias chasing headlines, recency bias extrapolating momentum, action bias generating churn.

The most useful concrete tool: before buying anything, list three specific reasons your thesis is wrong, then actively try to prove each one. Charlie Munger's inversion principle made operational. In a market where AI narratives are moving faster than fundamentals, that discipline is doing real work.

There's also a structural guardrail worth stealing: allocate a small, fixed amount into a separate account designated purely for active trading. It satisfies the compulsive urge to transact without touching the core portfolio. Given that leveraged ETF assets just grew 70% in four months, a lot of people appear to be missing this one.

The Pattern: Cycle Awareness Is the Actual Edge

What connects all three threads is cycle awareness — specifically, knowing which part of the cycle you're in. Galloway is describing late-cycle leverage behavior. Vishria is describing early-cycle infrastructure buildout. The Investing for Beginners crew is describing the behavioral failure mode that causes investors to conflate the two.

Geoffrey Hinton predicted in 2016 that radiologist training should stop because AI would outperform humans. He was technically right and practically wrong — he missed fragmented data, liability frameworks, and the Jevons paradox driving higher imaging volume. Vishria uses this as a template for why mass-unemployment predictions follow the same flawed reasoning. The technology works. The deployment path is never as clean as the prediction.

The AI trade isn't broken. But the leveraged, momentum-chasing version of it just demonstrated exactly what happens when cycle awareness goes missing.

---

---

This synthesis was AI-generated by SignalCast, which creates personalized podcast digests for the shows you listen to. Try it free →

Sources: Invest Like the Best with Patrick O'Shaughnessy, The Prof G Pod, Investing for Beginners · Fair use: all summaries link to original episodes