Goldman Sachs derivatives strategist Brian Garrett flags a clear market rotation into the second half of 2026:
investors are cutting exposure to the Magnificent Seven while piling into semiconductor stocks, the direct winners of AI capital expenditure.
The core divide hinges on returns versus spending. Markets reward firms that monetize AI infrastructure outlays (chipmakers) but grow wary of hyperscalers including Meta, Google, Microsoft and Amazon, who pour hundreds of billions into data center buildouts with delayed earnings payoffs. Garrett notes this trend has unfolded for months, driving a broader shift from asset-light large-cap tech toward asset-heavy industrial tech and pressuring Mag 7 valuations.
Until hyperscale operators deliver a visible upturn in profits, investors will keep underweighting mega-cap tech. Options pricing confirms rising downside risks: hedging costs for QQQ (Nasdaq 100 ETF) now sit far above small-cap equivalents, reflecting broad caution after the Mag 7 lagged the S&P 500 in recent months.

Semiconductors stand as the year’s top performing sector. AI buildouts created widespread memory and storage chip shortages, pushing product pricing higher and prompting manufacturers including Micron and Applied Materials to ramp capacity ahead of robust H2 demand. The global chip industry’s revenue nears $1 trillion this year.
ETF performance lays bare the stark divergence:
- Roundhill Memory ETF (DRAM): +141% since April launch
- VanEck Semiconductor ETF (SMH): +72%
- iShares Semiconductor ETF (SOXX): +99%
- Roundhill Magnificent Seven ETF: -7% from peak
- S&P 500: nearly +10%
Goldman Sachs expects this profit-spending split to remain the dominant positioning driver through end-2026.
