Billionaire Stanley Druckenmiller Dumped Broadcom and 11X'd His Fund's Stake in This Trillion-Dollar Dual-Industry Leader

Stanley Druckenmiller's Duquesne Family Office sold its entire 195,955-share Broadcom position in Q2 after the stock rose from roughly $300-$350 in Q1 to a June peak of $495, while increasing its Amazon holding by 495,800 shares, or 1,083%. The article cites potential AI-bubble risk as a factor behind Broadcom's exit, despite strong hyperscaler demand. Amazon's appeal rests on AWS, where growth reaccelerated to 37% in Q2 and annualized revenue reached $169 billion; Amazon trades at 11x projected 2027 cash flow per share.

Analysis

The actionable signal is not the reported manager activity itself: 13Fs are delayed, omit hedges, and a short holding period makes them especially poor indicators of current conviction. The more useful read is that AVGO has become a high-duration proxy for a narrow set of hyperscaler AI-capex budgets, while AMZN combines that same spend cycle with retail/logistics earnings leverage. If incremental AI infrastructure dollars migrate from merchant GPUs toward custom silicon and internal cloud capacity, AMZN can capture both AWS demand and lower unit compute costs; AVGO benefits from the buildout but remains more exposed to order timing and customer concentration.

Near term (days to 1-3 months), AVGO is vulnerable to any deceleration in AI networking/ASIC backlog conversion, not necessarily a collapse in end demand. The key issue is multiple compression: at elevated expectations, merely reaffirming guidance can be insufficient if management does not raise the next-quarter AI revenue outlook. Conversely, AMZN's valuation case depends on sustained AWS margin expansion rather than revenue growth alone; rising depreciation, power costs, and accelerated server refresh cycles could absorb much of the gross-profit benefit from AI services.

The contrarian view is that investors may be treating AMZN as a cheaper AI exposure without fully underwriting the capital-intensity required to defend AWS share against MSFT and GOOGL. A durable AWS growth reacceleration is bullish only if operating income scales faster than capex and lease commitments. For AVGO, a modest digestion phase in hyperscaler orders could be an entry opportunity rather than a structural break, provided custom-ASIC design wins broaden beyond its largest customers and networking attach rates remain intact.

Key Decisions for Investors

• Initiate a 1-3 month pair: long AMZN / short AVGO in equal beta-adjusted dollars. The thesis is relative multiple resilience if AI capex expectations normalize; target 10-15% relative outperformance. Exit if AVGO raises AI revenue guidance materially while AWS operating-margin guidance compresses or cloud growth underperforms MSFT/GCP.
• Do not trade on the disclosed 13F alone. Set an alert around the next AVGO earnings release for AI-semiconductor revenue, backlog, and top-customer concentration; a guide-down or weaker-than-expected sequential AI growth would support adding to the AVGO short, while a material guide-up falsifies it.
• For AMZN, add only after verifying that AWS segment operating income and free-cash-flow conversion improve despite higher capex. A 6-18 month long is attractive if AWS margins expand and retail fulfillment cost per unit continues falling; reduce exposure if capex-to-revenue rises without corresponding cloud-margin leverage.
• Use WMT as a hedge/watch comparator rather than a primary trade: if consumer discretionary demand weakens, AMZN retail margins may face greater pressure than implied by a cloud-led valuation narrative, while WMT's defensiveness can outperform.