13F Reveals New Signal: AI Has Not Faded, Wall Street Is Just Getting 'Pickier'
Title: 13F New Signal: AI Is Not Receding, Wall Street Just Became "Selective"
Analysis of 13F filings from the second quarter of 2026, which disclose institutional holdings, reveals a key trend: the AI investment theme persists, but Wall Street is now scrutinizing opportunities more selectively rather than chasing the sector broadly.
While nearly 44% of the 6,371 institutions analyzed reduced holdings in the "Magnificent Seven" tech giants, semiconductors saw net buying from 48% of institutions. This indicates the core AI infrastructure thesis remains intact. However, significant internal shifts are occurring as major funds prioritize "risk-reward" or "betting odds" over simple sector exposure.
Four prominent investors exemplify this new selectivity:
1. **Berkshire Hathaway** made a major new bet on Alphabet, valuing its strong cash flow and core businesses despite AI-related uncertainties.
2. **Tiger Global** reduced crowded mega-cap tech positions (e.g., Alphabet, NVIDIA) but increased exposure to other AI-related names like AMD and semiconductor manufacturing, rebalancing within the theme.
3. **Third Point** fully exited several first-wave AI winners (NVIDIA, Broadcom) to lock in gains, reallocating to names like Alphabet and Taiwan Semiconductor, and diversifying into media and industrials.
4. **Duquesne Family Office** (Stanley Druckenmiller) also sold some semiconductor holdings while buying others (e.g., Taiwan Semiconductor), focusing on individual companies' valuation and expectation gaps rather than the sector beta.
Key conclusions from the filings:
* **Alphabet is becoming a "divisive asset"** with significant institutional disagreement, representing a potential source of alpha.
* The **"buy any chip stock" phase is over**. Semiconductor investing is now an alpha game, requiring stock-specific analysis over blanket sector bets.
* **Non-AI assets are reappearing** in portfolios (e.g., airlines, homebuilders, media) as diversification and correlation-management tools.
In summary, the AI investment cycle has not ended, but the phase of easy, broad-based gains is concluding. The next phase will be defined by selective stock-picking and precise calculation of risk versus reward within the AI ecosystem.
marsbit11 min fa
