Here Are Billionaire Steven Cohen's 5 Biggest Stock Holdings

Point72's equity portfolio totals roughly $86 billion with more than 3,800 holdings and no single stock exceeding ~2.2% of assets. The five largest individual stock positions are Nvidia $1.87B (2.15%), Taiwan Semiconductor $1.41B (1.63%), Amazon $1.21B (1.40%), Microsoft $1.01B (1.16%), and Arista Networks $1.00B (1.14%). The portfolio shows a clear tilt toward AI infrastructure and large-cap tech, and including ETFs (e.g., QQQ) would further elevate passive/tech exposure among the top holdings.

Analysis

AI-infrastructure concentration creates asymmetric winners beyond the obvious GPU makers: advanced packaging vendors, HBM suppliers, optical/serdes makers and hyperscale networking (low-latency switches) will see demand growth that is both lumpy and high-margin. Expect meaningful lead-time-driven pricing pressure in HBM and advanced substrates over the next 6–12 months — that bottleneck amplifies near-term pricing power for incumbents even if unit growth moderates.

On the flip side, CPU-centric incumbents with legacy fabs and low 7nm+ utilization face margin compression as hyperscalers internalize AI stacks and push for accelerator diversity; Intel-style execution risk plus foundry competition can compress multiple valuation points in 6–18 months. Geopolitical/export-control shocks (China/Taiwan) and a hyperscaler capex pause are low-probability, high-impact tails that could unwind crowded positions within days to weeks via forced selling and options gamma mechanics.

Flows and options open interest make this market prone to fast directional moves: ETF concentration means a tech drawdown will mechanically hit the same large names and create cheap volatility spikes that are exploitable with defined-risk option structures. Over 12–24 months the structural bull case remains intact if wafer capacity and packaging investments lag demand, but that window is the battleground between pricing power and supply response.

The market consensus underweights the speed at which hyperscalers can disintermediate suppliers (internal chips, co-designed switches); that creates a scenario where current premium multiples get challenged before revenue shows weakness, making time-limited, convex option exposures superior to naked long equity positions.

Key Decisions for Investors

• Buy NVDA 3–6 month bull call spread: long ATM call / short +12–18% strike to capture near-term AI spend upside while capping cost. Rationale: asymmetric upside if HBM/accelerator tightness persists; max loss = premium, target return ~2–4x if NVDA rallies 20–40% in 3–6 months.
• Long ANET equity (size 2–4% portfolio) vs short INTC pair (equal notional) for 6–12 months: ANET benefits directly from hyperscaler networking upgrades while INTC faces execution/foundry risk. Target ANET outperformance of 15–25% vs INTC; hedge by sizing to neutral beta to broader market to limit macro risk.
• Buy TSM 9–12 month shares and sell 6–9 month covered calls to monetize elevated premium: play foundry pricing power and capacity tightness while collecting 4–8% annualized from calls. Risk: geopolitical disruption or rapid capacity adds could remove upside — keep 10–20% cash buffer.
• Purchase 1–3 month QQQ downside protection (buy puts or put spread) sized to cover 20–30% of tech exposure to guard against a volatility spike from an AI-capex pause or export-control shock. Cost is insurance premium; benefit is rapid drawdown protection in an ETF-driven selloff.