What needs to happen for current market levels to make sense
Good question to ask separately, because it forces the argument to be precise rather than just vibes-based. Here's what the math actually demands.
The Starting Point
The S&P 500's trailing PE is around 29, against a long-run historical average of roughly 19.7. The Shiller CAPE — which smooths earnings over 10 years to remove cyclical distortion — sits at about 39.6 for May 2026, one of the highest readings in modern history outside of 1999-2000. The forward PE, using estimated future earnings, is around 21. Zacks + 3
That gap between trailing (~29) and forward (~21) is itself a signal: the market is essentially betting hard on an earnings surge. For that bet to pay off, several things need to go right simultaneously.
Condition 1: Earnings growth has to actually land at ~12%+ for multiple years
Goldman Sachs projects EPS growth of 12% in 2026 and 10% in 2027. That's not outlandish by itself, but it needs to hold up in an environment of elevated oil prices, elevated bond yields, and potential tariff drag — all of which compress margins. If earnings disappoint even modestly, the forward PE re-rates back toward the trailing PE, and prices fall to compensate. Goldman Sachs
Condition 2: "Higher for longer" rates have to be survivable
At a 21x forward PE, you're paying $21 for every $1 of expected earnings. The risk-free rate (10-year Treasury) is currently well above 4%. That matters because valuation multiples are essentially inverse discount rates — when the "safe" return goes up, the premium you pay for risky future earnings should compress. For a 21x multiple to hold, the market needs to believe either that rates will fall meaningfully, or that earnings growth is so reliable that the premium is still justified. Right now, it's betting on both.
Condition 3: AI has to deliver actual productivity gains — at scale, soon
This is the biggest wild card. The bull case embedded in current prices isn't just that AI companies make money — it's that AI raises the productivity ceiling of the entire economy, justifying both higher earnings and higher multiples across most sectors. That's a very large assumption. Technology historically does deliver productivity gains, but with long and variable lags (the internet's productivity boom didn't show up clearly until the mid-2000s, a decade after the infrastructure was built).
Condition 4: No recession
This sounds obvious but it's load-bearing. Goldman cites healthy economic and revenue growth as foundational to its bull case. A recession wouldn't just cut earnings — it would compress the multiple at the same time, creating a double hit. At current valuations, even a mild recession could produce the 30-40% drawdown Niles describes, simply through the mechanical math of earnings falling while PE normalizes. Goldman Sachs
Condition 5: Concentration has to remain a feature, not a bug
A significant portion of S&P 500 returns are driven by a handful of mega-cap names. For the index to hold up, those companies need to keep growing into their valuations. If even one or two of them disappoint — on earnings, regulation, or competitive disruption — the index-level math deteriorates quickly.
The honest summary
Current market levels aren't insane if you believe in a specific scenario: no recession, sustained double-digit earnings growth, AI productivity materializing within 2-3 years, and rates drifting lower. That scenario is possible. But it requires nearly everything to go right. The market isn't pricing in risk — it's pricing in a best case. That's not the same as the market being wrong; it just means the margin for error is very thin.