Goldman Sachs notes shift in fund positioning towards semis and away from software

Goldman Sachs says hedge funds and mutual funds shifted further toward semiconductors and away from software at the start of Q2 2026, with mutual funds holding their lowest software exposure since at least 2012 and semis at the largest tilt versus software since 2012. Hedge funds added to Lam Research, Applied Materials, and ASML, while mutual funds added Intel and SiTime; Microsoft saw one of the biggest net-share reductions. The positioning data reinforces the market's AI-driven preference for semiconductor exposure over legacy software, though the article is mainly flow commentary rather than a company-specific catalyst.

Analysis

Positioning is now making the AI trade self-reinforcing: flows are concentrating into the semiconductor toolchain just as software multiples are being rerated lower for structurally weaker earnings durability. The first-order winner is not just chip designers, but the capital equipment layer where order books can remain elevated even if end-demand cools; that helps explain why LRCX, AMAT, and ASML screen better than the more crowded mega-cap AI beneficiaries on a 3-6 month horizon. The second-order loser is the enterprise software basket, where AI is compressing both growth expectations and defensibility simultaneously, creating a negative feedback loop in fund ownership.

The sharp reduction in MSFT exposure is more telling than the broader software underweight. That kind of de-risking usually signals investors are no longer paying up for “platform tax” durability and are instead demanding proof that AI monetization offsets cannibalization risk; until that proof arrives, MSFT can lag even if fundamentals remain high quality. META and AAPL receiving net additions suggests capital is rotating toward names where AI can be monetized through ads, devices, or ecosystem control rather than through seat-based software pricing.

The biggest contrarian risk is overcrowding in semis: when mutual funds and hedge funds are both stretched long the same industrial-capex beneficiaries, any disappointment in AI capex or export restrictions can trigger a fast de-grossing. INTC and SITM are more idiosyncratic than the rest of the group; INTC’s ownership increase looks like a catch-up/value-reversion expression, while SITM benefits from a more durable edge-silicon narrative but remains sensitive to handset and industrial demand. A policy shock that lifts rates or cools tech multiples would hit software first, but it would also expose the semis’ implied growth assumptions quickly, so the trade works best with disciplined time stops rather than just fundamental conviction.

Key Decisions for Investors

• Long LRCX / AMAT / ASML basket vs short IGV or a software-heavy peer basket for 3-6 months; thesis is flows plus earnings revision divergence, with semis still supported unless AI capex slows materially.
• Fade MSFT on rallies into earnings via a put spread or a small tactical short for 4-8 weeks; risk/reward is attractive if the market continues to discount software moat erosion before monetization proof emerges.
• Add META/AAPL on pullbacks as the cleaner AI monetization expressions over 1-2 quarters; these names have better upside-to-drawdown profiles than pure software because AI can lift existing distribution rather than threaten it.
• Use INTC as a relative-value long only versus software or weaker legacy infrastructure names, not as a standalone outright long; the setup is more re-rating than secular growth and should be treated as a trade, not a thesis.
• If semis gap higher again, trim into strength and rotate into the most underowned AI beneficiaries with operating leverage, because the risk/reward worsens quickly when positioning becomes consensus-long.