Hedge fund tech positions hover near record highs, Goldman Sachs says

Hedge funds bought technology stocks at the fastest pace in nearly three months, with buying led by North America and Asia emerging markets. Positions were concentrated in semiconductor-related manufacturers and software, while communications equipment and IT services were sold. The note highlights record-high bets on global information technology stocks and the largest relative tech exposure versus the MSCI World index in over five years.

Analysis

The flow signal matters more than the headline: when hedge funds crowd into semis/software while cutting defensive underweights, the market stops pricing earnings and starts pricing positioning. That creates a short-term “air pocket” risk for anything with crowded growth ownership, because marginal buyers are no longer fundamental investors but systematic and momentum-linked accounts that can reverse quickly if rates back up or macro data cools AI capex enthusiasm.

The second-order winner is not just the obvious semiconductor leaders, but the broader AI supply chain where order visibility can stay elevated even if end-demand is noisy: foundry, advanced packaging, high-bandwidth memory, and test equipment tend to benefit later and with less valuation fragility than the highest-multiple software names. By contrast, communications gear and IT services are the likely funding sources for this trade; if the AI theme broadens again, those laggards should continue to underperform as budget is reallocated toward compute rather than implementation.

The main risk is that this positioning becomes a consensus trade into a macro wobble. If long-end yields rise or any export-control / China policy shock hits, semis can de-rate 10-15% in days even with unchanged demand, because the tape is already crowded. Over a 1-3 month horizon, the trade is still constructive, but the edge is increasingly in structure and relative value rather than outright beta.

The contrarian read is that record-long exposure often marks a late-cycle phase where good news is already owned. That does not argue against AI, but it does argue for owning the franchises with operating leverage and balance-sheet support while fading the most expensive software/infra names that need perfect execution to justify current multiples. In other words, the trade is becoming less about being long AI and more about being selective within AI.

Key Decisions for Investors

• Long SMCI vs. a basket of IT services/communications equipment for 4-8 weeks; the idea is to own the most direct beneficiary of incremental AI capex while shorting the funding source of the trade. Use tight risk controls because SMCI’s valuation and volatility can gap on any demand-delay headline.
• Add tactical long GS on a 1-2 month horizon as a cleaner expression of the flow/financing pick-up from elevated equity turnover and AI-linked issuance activity. Risk/reward is better as a moderate beta proxy than as a pure AI expression, but size should be smaller than semis because upside is more derivative than direct.
• Initiate a pairs trade: long SMH / short IGV or an equal-weight software basket over the next 4-6 weeks. This captures the second-order rotation from AI story stocks into the compute stack, with better odds if rates remain stable and earnings revisions stay positive.
• For higher-conviction accounts, buy 1-3 month call spreads in SMCI rather than outright calls. Volatility is rich and crowded positioning raises gap-risk; spreads preserve upside if the AI trade extends while limiting premium bleed if the crowded long unwinds.
• Trim or hedge any broad tech overexposure if MSCI World Information Technology exposure is already overweight versus benchmark. The risk/reward has shifted from accumulation to defense: keep core AI winners, but reduce beta where performance depends on continued crowding rather than fresh fundamental inflection.