Situational Awareness wasn't the only one that had a bad July. Hedge funds took a historical hit last month

Technology-focused hedge funds fell 10.2% in July (worst month on record), versus multi-strategy funds down 2.3%, reinforcing fears that hedge funds have limited capacity to re-lever and fund the post–July-rout tech rally. JPMorgan warns that reduced hedge-fund risk appetite—and potential prime-broker lending pullbacks—could make technology investment capacity “structurally more limited” going forward, increasing the rally’s dependence on retail flows and leveraged ETFs/option buying. Momentum and semiconductor ETFs were hit hard (MTUM -12.6%, DRAM ~-32%, SMH ~-18%, SOXX -21%), while Sandisk shares slid ~46% in July and dropped another ~5% after disappointing revenue guidance, suggesting fundamentals are still being questioned.

Analysis

The key market mechanism is not the July drawdown itself, but the loss of institutional marginal bid in the most crowded AI complex. When hedge funds are forced to de-gross and prime brokers tighten financing, rallies become more dependent on retail call flow and levered ETF churn, which usually means higher intraday upside but weaker multi-week follow-through. That is most negative for names whose valuation still depends on external capital and narrative momentum rather than self-funding cash generation: CRWV, NBIS, SHAZ, and to a lesser extent BE.

SNDK looks like the cleanest expression of that vulnerability because it combines cyclical fundamentals with positioning risk; if memory pricing softens while risk appetite fades, multiple compression can accelerate faster than earnings revisions. A second-order effect is that AI-adjacent power and infrastructure names can de-rate even if their operating story is intact, simply because the market stops paying for the long-dated capex option value. JPM is a relative winner only in a narrow sense: more volatility and financing demand can support prime-broker economics, but the bigger upside is balance-sheet optionality, not directional equity beta.

The contrarian risk is that hyperscaler capex validation may have reset expectations enough to trigger a fast re-leveraging of the trade. If SMH/DRAM regain prior breakout levels while retail options activity stays elevated, the current unwind can turn into a squeeze rather than a multi-month reset. What would falsify the bearish positioning thesis is evidence that hedge-fund leverage is stabilizing or rising again, or another round of earnings confirming AI-related spend is converting into cash returns rather than circular demand.

Key Decisions for Investors

• Short SNDK on any bounce toward the prior breakdown area; target 10-15% relative underperformance versus SMH over 4-8 weeks, with a tight stop if guidance improves or memory pricing turns up.
• Pair trade: long JPM / short CRWV for 1-3 months as a financing-stress hedge. Thesis works if prime-broker conditions tighten and crowded AI names remain supply-heavy; risk is that CRWV squeezes on renewed retail momentum.
• Sell rallies in NBIS and SHAZ via put spreads into strength rather than chasing weakness. This is a volatility/positioning trade, not a thesis on near-term fundamentals; best entry is after a 1-2 day reflexive bounce.
• No immediate trade in BE; keep it on watch as an AI-power second-order beneficiary. Revisit only if data center capex is reaffirmed and the stock can hold above its post-rout moving average for several sessions.
• Use SMH/DRAM as a sentiment gauge: if either ETF fails to reclaim the July breakdown level within 2-3 weeks, add to defensive shorts; if both reclaim it on strong breadth, cover crowded shorts quickly.