It Only Happens at Bottoms | Andy Constan on the Options Extreme That Showed Up at the Highs

Episode Summary

Executive Summary: Andy Constant argues the AI/semiconductor surge has become a speculative bubble driven by extraordinary capex, low correlation, and unusual options-market behavior (bid calls, rich single-name vol, weak hedging). He says the economy is being supported by AI capex and consumer dissaving, but there isn’t enough “pie” for all the promised earnings claims. He also sees limited near-term Fed support for equities, a more balanced Warshaw regime, and a potentially important tariff reset.

Main Topics: Options-market extremes in AI/semis (Priority: 5/5): Andy explains that unusual call demand, elevated single-stock volatility, and low correlation among stocks signaled a speculative frenzy in AI infrastructure and semiconductor names before the pullback. AI boom and the “pie” problem (Priority: 5/5): He argues that AI can be transformative, but current earnings expectations across companies exceed the economic pie available, meaning not every firm can win the returns investors are pricing in. Bubble regime and market dispersion (Priority: 4/5): He frames the current environment as a bubble with parabolic price moves, concentrated winners, and a broad market that looks flat even as individual names swing violently. Economic support from capex and dissaving (Priority: 4/5): The economy is okay for now because it is being propped up by capital expenditures and consumer dissaving, though dissaving cannot continue indefinitely. Fed policy, equities, and the Warshaw approach (Priority: 4/5): Constant doubts the Fed will buy equities except in an extreme crisis, but he is receptive to a smaller balance sheet and lower policy rates as a better policy mix than today. Tariff deadline and trade policy reset (Priority: 3/5): He highlights that key tariff measures are expiring soon and expects near-term policy news that could materially affect growth, inflation, and markets.

Key Arguments: Call-option pricing in AI/semiconductor names became unusually rich, suggesting a market that was aggressively chasing upside rather than hedging downside. Single-stock realized and implied volatility rose sharply, especially in semiconductor-adjacent names, which is consistent with winner/loser dispersion inside a bubble. A bubble is visible not just in valuation multiples, but in earnings expectations that assume explosive growth can continue indefinitely. The AI capex boom is financing a narrow set of companies, but the aggregate economy may not generate enough incremental GDP to justify all of the promised future cash flows. Even under aggressive assumptions about productivity, population, and corporate share gains, the model still shows insufficient “pie” to meet current earnings expectations. Consumer dissaving and heavy AI-related capex are propping up the economy now, but dissaving has a natural limit and capex depends on market funding capacity. The Fed is unlikely to buy equities in normal conditions, but in a severe downturn it would likely cut rates and expand QE, potentially including ETFs. A smaller Fed balance sheet plus lower short-term rates would, in his view, modestly rebalance asset prices and reduce some asset-holder advantage. Tariffs are a near-term wildcard because some major measures are set to expire and need replacement, making policy changes likely soon.

Data Points: Episode status: Fourth episode of First Principles - The hosts introduce the discussion as the show’s fourth installment with Andy Constant. Capital expenditures: Massive - Andy says AI infrastructure spending is a major economic driver and requires financing from markets. Concentrated winners: About 3 companies - He says a few firms dominate memory supply and a few others dominate compute chips. Raw silicon makers: “One or two two-ish, really just one company” - He describes a highly concentrated chain for raw silicon. Fab five: Five or so companies - His term for companies necessary for clean-room builds. Intel stock move: 3x to 4x - He cites Intel’s parabolic rise as an example of extreme semiconductor moves. Correlation among stocks: Very low - He says individual-stock correlation is unusually low while the index is relatively flat. Single-stock volatility: Doubled or tripled in some cases - He notes realized and implied vol in semiconductor names have surged. Index volatility: Relatively elevated - He says index option volatility remains high versus realized index volatility. Interest rates: Fed funds could go to zero in crisis - He says a severe downturn would force aggressive easing. U.S. stock ownership: Roughly 58% of Americans - Cited from Eric Balchunas’ comment on market centrality to household wealth. Potential stock ownership: Could approach 70% - Balchunas’ cited estimate if Trump accounts expand participation. U.S. debt: $39 trillion - Andy references high debt as a reason the U.S. is more financialized. COVID stimulus: About $6 trillion - He cites fiscal spending financed by the Fed during COVID. Tariff deadline: July 24 - He says Section 122 tariffs are expiring and need replacement. Current date reference: July 17 - He notes there is roughly a week left for the administration to outline new tariffs. Tariff impact estimate: $100 billion - He estimates the tariff regime change could move the deficit by this amount. GDP impact estimate: 25 to 30 bps - He says the tariff change could affect GDP by this magnitude.

Pivotal Quotes: "There’s a bull market for calls and a bear market for puts." — Andy Constant: He summarizes the unusual options-market behavior around AI and semiconductor names. "I think there’s not enough pie not even close to enough pie." — Andy Constant: He describes why aggregate earnings expectations across AI beneficiaries may be impossible to satisfy. "Talk to me when stocks are 80% lower. Then the Fed may buy some ETFs." — Andy Constant: He rejects the idea that the Fed is likely to support equities at current levels.

Implications: The episode suggests investors should treat AI semis as a crowded, volatile trade, not a one-way story. Near-term risks include tariff changes and continued funding pressure on capex; longer term, AI may still be productive, but many current earnings expectations look too optimistic.