The Perpetual Momentum Machine

The current momentum trade is effectively a highly concentrated AI bet whose attractive earnings narrative is paired with stretched valuations and fragile diversification.

• Momentum typically selects the best-performing stocks over six or twelve months, and MAN Numeric found an approximately 0.86 historical correlation between price momentum and earnings momentum through January 2025.
• The AI-led rally has accelerated sharply: the S&P 500 Momentum Index rose 31.7% in April and May, its largest two-month move in 32 years; the PHLX Semiconductor Index gained 89% year to date; and technology stocks generated 85% of S&P 500 gains.
• Price momentum temporarily moved ahead of earnings momentum as investors bought AI-related stocks regardless of earnings upgrades. Hyperscaler spending is approaching $700 billion this year, while the S&P 500 forward P/E is 21.8x and 2027 EPS estimates have risen 11.6% in six months.
• The earnings assumptions are especially demanding outside the Magnificent 7: aggregate S&P 500 margins are expected to rise from 16% to 22% by the end of 2028, while margins for the other 493 companies would need to climb from 14% to 20%, far above their historical trend.
• Diversification may be illusory because AI-linked stocks represent 45% of the S&P 500, TSMC, Samsung and Hynix represent 27% of the MSCI Emerging Markets Index, US retail investors have a record 45% equity allocation, and hedge-fund gross leverage is 3.2x.

Takeaway: Investors should not rely on regional, stylistic or strategy diversification to protect them from an AI-led reversal; the article favours businesses with predictable earnings growth independent of the AI cycle, while acknowledging that this may require tolerating near-term underperformance.