Santoli: Why investors should be on high alert heading into September

The market is priced for a soft landing with unusually little volatility, but the bigger issue is fragility of leadership rather than absolute valuation. When breadth is this narrow and a handful of names are carrying a disproportionate share of forward earnings growth, index stability can mask a growing single-factor risk: any small disappointment in AI capex, margins, or guidance can force multiple compression across the whole growth complex, not just one stock.

Rates are the cleaner near-term catalyst than seasonality. A 10-year yield hovering near 5% with a VIX below 15 is an uncomfortable mix for crowded long-duration equity exposures: it does not automatically break the tape, but it raises the odds of a sharp de-grossing if a CPI or labor print confirms sticky inflation. That setup favors quality balance sheets and cash-return names over cyclical beta, while exposing consumer-facing and rate-sensitive stocks to another round of estimate cuts if real activity softens.

The contrarian point is that the consensus is likely over-fixated on September as a calendar effect and underweight the regime issue: higher-for-longer yields plus concentrated earnings leadership. If yields stabilize and AI spending remains exponential, the market can keep grinding higher despite weak breadth. But if the next leg is driven by volatility in rates rather than earnings, the first casualties will be the expensive momentum sleeves and lower-quality financials, not necessarily the index.

For NVDA, the problem is not near-term demand but proof of durability; until the market believes earnings power is recurring rather than hit-driven, it will keep capping the multiple even on strong prints. AAPL is the cleaner defensive AI beneficiary because it offers cash-flow visibility and optionality without requiring perfect execution. JPM screens as the better way to express a higher-volatility, higher-rate regime versus C, which has less margin for error if risk appetite fades.