Billionaire Stanley Druckenmiller Sells Broadcom Stock and Buys an AI Stock Up 1,000% Since Early 2025

Stanley Druckenmiller sold Broadcom and initiated a position in Sandisk in Q3, swapping one semiconductor exposure for another. Broadcom, with ~75% share in AI ASICs and AI-related revenue rising 65% to $20 billion in 2025, has Wall Street adjusted EPS growth of ~43% through 2027 and trades at ~51x forward earnings (current price $343, median target $461, ~34% upside). Sandisk, a vertically integrated NAND flash supplier that gained share in H1 2025 and is testing enterprise SSDs with hyperscalers, has much higher forward multiples (about 170x) despite consensus adjusted EPS growth of ~79% through fiscal 2029; the stock rose ~7x since Druckenmiller's Q3 buy (current ~$415 vs Q3 average $58) and carries a median target implying ~26% downside.

Analysis

Market structure: Broadcom (AVGO) is the clear winner in AI ASICs (≈75% share) and networking silicon, giving it pricing power into hyperscaler spend; its 51x forward P/E implies growth priced in but not extreme versus a Street CAGR ~43% to 2027 and a median target of $461 (+34%). Sandisk (SNDK) benefits from rising SSD share and vertical integration but trades at ≈170x forward earnings after a 7x post-Q3 run and looks susceptible to NAND ASP cycles; HDD names (WDC) and commodity NAND suppliers are the likely losers if SSD adoption plateaus or prices collapse.

Risk assessment: Tail risks include export/regulatory curbs on AI chips, hyperscaler vertical integration displacing third-party ASICs, and a NAND oversupply causing ASP declines of 20–30% within 12 months. Near-term (days–weeks) moves will be sentiment-driven around earnings; medium-term (3–12 months) depends on hyperscaler testing converting to contracts; long-term (2–3 years) hinges on data-center AI capex reaching I/O Fund/JPMorgan projections (~3x–5x vs. 2025). Hidden dependency: revenue concentration to a handful of hyperscalers (Google, Meta, OpenAI/Anthropic).

Trade implications: Tactical play is to accumulate AVGO on pullbacks (<$320) sized 2–4% of portfolio with 12–18 month target $461 and a stop ~18% below cost. Reduce/exploit SNDK by trimming longs or initiating a small (1–2%) short or buying long-dated puts (Jan 2027 strike ~300–350) given stretched 170x multiple and downside to $307 median target. Pair trade (long AVGO / short SNDK) captures relative quality in AI stack and should be rebalanced quarterly; overweight networking and semiconductor capex suppliers, underweight commodity NAND.

Contrarian angles: Consensus overlooks single-customer concentration at Broadcom and execution risk converting hyperscaler tests into volume; AVGO could re-rate higher if AI spend accelerates beyond JPMorgan’s $110B by 2027, but a single large contract loss would be acute. Conversely, SNDK’s momentum trade may be overdone—historical NAND cycles (2016–2019) show rapid mean reversion—so price action is vulnerable to a 30%+ drawdown if ASPs soften or Kioxia JV capex expands capacity. Monitor hyperscaler procurement cadence and NAND ASPs as primary catalysts.

Key Decisions for Investors

• Establish a 2–4% long position in AVGO on a pullback to <$320; add another 2% if price drops below $285. Target $461 in 12–18 months and use a stop-loss ~18% below your average entry to limit downside.
• Reduce or exit any SNDK longs immediately; if seeking opportunistic short exposure, initiate a 1–2% portfolio short or purchase Jan 2027 puts (strike 300–350) targeting $307 median price within 3–12 months. Place stop-loss for shorts at +20% above entry.
• Implement a pair trade: equal-dollar long AVGO / short SNDK sized to 2% net exposure to express conviction in ASIC/networking vs. commodity NAND; rebalance quarterly and widen if NAND ASPs decline >15% QoQ.
• Use options to enhance yield/hedge: sell AVGO Oct 2026 covered calls at ~420 if long to collect premium; buy SNDK Jan 2027 puts (350 strike) as a tail hedge if worried about a NAND bust within 12–18 months.
• Monitor three specific catalysts over next 90–180 days: (1) Broadcom quarterly AI revenue cadence and customer disclosures; (2) SNDK hyperscaler testing updates and contract wins; (3) NAND ASP movements—increase short/hedge if ASPs fall >15% QoQ.