James Mackintosh's Streetwise column on B1 today titled "Why The S&P Rally Rests On One-Offs" is the contrarian read that every Capital Wealth client needs to hear before we extrapolate the last three weeks into an AI euphoria year. Mackintosh's argument is disciplined and empirical: a huge chunk of 2026's S&P earnings growth is one-time-in-nature, and markets are treating it as recurring.
Mackintosh names three specific one-offs:
- Tariff-driven pricing power. S&P 500 companies have passed through tariffs into prices. That lifts nominal revenue but it's a one-time re-pricing, not a structural growth rate.
- Hyperscaler AI capex. MSFT/GOOGL/META/AMZN together are spending roughly $400 billion on AI infrastructure in 2026. That cash is landing as revenue at NVDA/AVGO/VRT/etc., and inflating S&P EPS growth by an estimated 2.5 points. It is not a steady-state capex number.
- Dollar weakness. A soft dollar is boosting the foreign earnings of multinationals this quarter. Also not a growth rate.
Our Read — Why This Reinforces Theme Discipline
Mackintosh is right. And the answer is not to go to cash — it is to be in the specific themes where the earnings power is real: Energy (18%), Defense (16%), Power/Infra (10%), Gold (12%), Nuclear (5%). Those sit on structural, not cyclical, tailwinds. The AI/Cyber/Data sleeve (20%) is exposed to Mackintosh's critique but is balanced by the picks-and-shovels subset (NVDA/AVGO/VRT) that directly benefits from the very capex he calls a one-off.
Three Defensive Moves If Mackintosh Is Right
SGOV, BIL, JPST
2Y at 3.7% means T-bill yields are still compelling. We are rotating 3-5% of portfolios here as dry powder.
GLD, IAU, NEM, AEM
$4,857 ATH. Defensive hedge against both a reflation overshoot and an earnings disappointment.
PG, KO, MCD, WM, VZ
If earnings disappoint and the Fed can't cut fast, boring consumer staples outperform the multiple-compression trade.
Book Alignment — Mackintosh Discipline
We run our defensive sleeve at 8% — staples + short-duration Treasuries. Mackintosh's column is the reason we don't push that lower even when the tape is ripping.