PodcastThe Monetary Matters NetworkAndy Constan

Why Andy Constan Says The AI Bubble is in Earnings, Not Price

Policymakers' fiscal and monetary missteps—combined with inflationary fiscal measures and geopolitical strain—have kept inflationary pressures high and, absent aggressive deficit reversal, support a higher-for-longer rate outlook. Andy frames AI as a compute-driven advance whose economic payoff depends on customers capturing real productivity gains, creating a mismatch between modest GDP growth (roughly $1.5 trillion) and outsized earnings expectations concentrated in a few firms. That concentration underpins an 'earnings' bubble—reflected in mega IPO risk, true-believer investor behavior, and call-heavy option skew—which elevates downside and absorption risks and favors pragmatic hedges like covered collars.

Market ValuationsArtificial IntelligenceMarket BreadthInvestor SentimentMonetary PolicyFiscal Policy

Preview · 9 more points in the app

Read the Full Summary

Key Takeaways

Persistent fiscal actions (e.g., Inflation Reduction Act spending, gasoline tax-cut proposals) combined with a failure to decisively reverse deficits are keeping inflationary pressure alive, and a K-shaped distribution—where top-end consumers sustain spending despite stagnant wages—helps entrench a 'higher-for-longer' interest-rate outlook.

Market consensus is pricing an outsized earnings allocation to AI and semiconductor firms (roughly 62.5% of incremental S&P earnings), but with expected global GDP expansion of only about $1.5 trillion there's a finite 'pie'—making those concentrated earnings expectations materially at risk.

AI progress is heavily compute-driven with concentrated capex in semiconductors and hyperscalers, yet monetization depends on end customers capturing productivity gains—a shift that 'has not yet happened'—so large compute investments may not produce near-term returns.

A cohort of 'true-believer' AI investors expecting another two to four years of upside can sustain elevated valuations, but history and the guest warn that a bubble unwind could be catastrophic (70–80% peak-to-trough moves cited) and behavioral mistakes like failing to take profits amplify losses.

GEO-POLITICS

Policy Failure — The speaker says monetary and fiscal policymakers 'dropped the ball' after months of inflation above target and argues recent fiscal moves (e.g., Inflation Reduction Act spending proposals, gasoline tax-cut proposals), tariffs, and geopolitical conflict are inflationary. He contends a genuine cooldown requires aggressive deficit reversal via spending cuts or tariffs rather than tax cuts, but sees no political appetite for that.

“The policy makers, the Fed has completely dropped the ball. Never really took in inflation by the horn by the horns and wrestled it to the ground…”—

TECHNOLOGY

Compute-Driven AI — Andy frames AI as a long, compute-driven evolution from earlier neural-net approaches to today's frontier models, but says the economic payoff depends on customers realizing productivity gains. Heavy compute-led capex concentrated in semiconductors, frontier models, and hyperscalers therefore may not guarantee near-term returns if purchasers can't afford or capture those gains.

“It's been a slow grind from regressions to neural networks to machine learning all driven by compute and the imp the rapid increase in compute that has led us to…”—

ECONOMY

GDP-Earnings Gap — Andy highlights a numeric mismatch: roughly $1.5 trillion of expected GDP growth versus consensus forecasts that allocate a disproportionately large incremental share of S&P earnings to public AI stocks, raising the risk that those earnings expectations are unsustainable given a finite global GDP 'pie.'

“Currently, there's a GDP growth year-over-year that's expected to be about $200 billion dollars of new GDP. Sorry, about a billion and a half new GDP…”—

MARKETS

Earnings Bubble — Andy argues the bubble is in earnings rather than prices: consensus expects a very large share of S&P earnings growth to accrue to AI and semiconductor firms—he cites 62.5%—and if those concentrated earnings don't materialize, AI-heavy stocks and the broader index could suffer steep declines.

“What I think it's driven by is extraordinary unsustainable E. It's not the P that's the bubble. It's the E.”—

True Believers — Many AI and semiconductor investors describe themselves as 'true believers' who accept bubble risk but expect the boom to last two to four years, a mindset that can keep valuations elevated despite warning signs.

“Some of the people think yeah this is a bubble this is a total boom but it's going to last two to four more years and immense money is going…”—

Mega IPOs — A very small set of mega IPOs—examples cited include SpaceX and two frontier-model companies—could collectively rival the proceeds of the late-1990s wave of 400 IPOs, further concentrating corporate scale and claims on global GDP.

“We're going to have three IPOs in the three, you know, SpaceX and the two frontier model companies that in total, even adjusted for GDP or, you know, inflation or…”—

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal