Raoul Pal and Andreas Steno discuss the difficulty of investing amid the AI S-curve — where LLM capabilities double roughly every 4-5 months and the market is plagued by rolling sector drawdowns as investors struggle to separate signal from noise. Steno highlights the paradox of the EM/AI trade: investors flee to emerging markets to escape AI risk, but EM indices are heavily concentrated in Korean and Taiwanese semiconductor names (Samsung, SK Hynix, TSMC) that are on the receiving end of AI capex. The core tension: markets are terrified of the AI "kiss of death" hitting individual sectors yet continue to buy the very semiconductor manufacturers that benefit from sustained capex.
Watch on YouTube ›Get the market thesis, key claims, assets, contradictions, and follow-up questions from any financial video — then unlock a version personalized to your portfolio, watchlist, and favorite speakers.
Raoul Pal opens with a provocative framing: within five years, we will have AGI brains inside humanoid robots — entities smarter than humans by 5-10x and physically superior. He argues nobody has internalized what this means. Andreas Steno pivots this to the investment problem: the LLM S-curve is so steep and early that extrapolation is impossible, analogous to early COVID modeling where Imperial College projected catastrophic outcomes from an exponential curve whose slope was unknowable. Steno's central observation is that whenever AI touches a sector, the market's reflex is to price the left-tail — the "AI kiss of death." He cites trucking (a white paper suggesting 300% output gains without added costs triggered a 25% sector wipeout overnight) and SaaS as examples. …
Tactically cautious but not outright bearish: rolling sector drawdowns create buyable dislocations in AI-adjacent names that get repriced on unclear signals, but timing these rebounds is difficult. The dispersion trade (short index volatility, long single-stock volatility) appears to be the dominant near-term regime.
The AI capex cycle continues to benefit semiconductor manufacturers (TSMC, Samsung, SK Hynix) and data-center infrastructure, while the capex-payers (hyperscalers, software) face ongoing ROI scrutiny. The EM/semis trade has further room as long as capex budgets hold, but the first sign of capex cuts would reverse it sharply.
AGI deployment represents a structural regime shift that will fundamentally reprice labor, capital allocation, and sector economics. Markets currently have no framework for this and are pricing it via erratic sector-level repricings — this pattern will persist for years as the S-curve plays out.
Within the next 5 years, we will have AGI brains in humanoid figures that are smarter than humans by 5-10x, stronger, and more adaptable.
The speaker asserts this as an inevitable development that most haven't figured out yet, without citing specific evidence.
LLM capabilities roughly double relative to competing with a software engineer every 4-5 months, and we are very early in that S-curve.
Speaker references algorithmic scaling trends but acknowledges the difficulty of extrapolating the S-curve.
The S&P 500 has never seen this many stocks with drawdowns bigger than 7% in single-day trading periods without a complete crash in the index.
Speaker makes a factual historical claim about market breadth and drawdown patterns to illustrate unusual sector rotation dynamics.
How does the rapid advancement of AI, particularly AGI in humanoid form, make the investment environment tricky?
The guest explains that LLMs are early in their S-curve, roughly doubling their capabilities every 4-5 months in competing with software engineers. He uses an analogy to COVID where extrapolation was impossible early on. He notes that when sectors face the 'AI kiss of death,' investors extrapolate harshly, causing sudden drawdowns like in trucking (25% overnight after a white paper) and SaaS. The S&P 500 has seen an unprecedented number of stocks with 7%+ daily drawdowns without a full index crash, showing the market cannot distinguish signal from noise.
Unlock the full claims, asset map, scores, related transcripts, follow-up questions, and AI chat — shaped around your portfolio, watchlist, favorite speakers, and risks.
