Could Broadcom Become the Next Nvidia?

Broadcom reported AI revenue up >100% to $8.4B last quarter and expects AI revenue to exceed $10B next period, with chips-only revenue potentially topping $100B in 2027. Networking made ~33% of AI revenue (forecast to rise to ~40%); major customer orders include $21B from Anthropic plus deals with OpenAI, Meta and Alphabet. The company highlights strong demand for Tomahawk 6 switches and XPU partnerships with six major customers; Broadcom trades at ~29x forward earnings with a market cap near $1.5T, implying potential upside versus larger peers like Nvidia.

Analysis

Broadcom’s move into bespoke AI silicon and upgraded switching fabrics materially shifts the bargaining dynamic between hyperscalers and their suppliers: bespoke XPUs raise customer switching costs and create multi-year revenue visibility for the vendor, but they also concentrate commercial risk into a handful of design wins. That concentration amplifies headline risk — a single large program delay or a foundry capacity reallocation can produce multi-quarter revenue volatility despite otherwise strong secular demand.

On the supply side, the transition from mostly firmware/ASIC refresh cycles to high-volume, process-node-dependent chip manufacturing means Broadcom is now hostage to foundry allocation and substrate capacity; price is no longer the only constraint, lead-times and yield curves will determine throughput. This also creates a practical choke-point for competitors: incumbents with established switch ecosystems gain the luxury of gradual margin improvement, while newcomers face both silicon and software barriers to entry.

Nvidia’s position is complementary not adversarial in many AI stacks — GPUs and customer-specific accelerators can coexist across training/inference and cost-performance bands — so Broadcom’s ascent is unlikely to cannibalize total GPU TAM immediately, but it will compress gross-margin trajectories for cloud buyers and force more nuanced workload placement. Finally, regulatory and contract-concentration risks are underappreciated: rapid revenue scaling combined with a small customer base is the exact profile that draws antitrust, export-control, or procurement-scrutiny, any of which could truncate the current multiple expansion story within 12–24 months.

Key Decisions for Investors

• Long AVGO via a two-year call-spread (buy Jan‑2028 LEAP call, partially finance by selling a nearer-term call). Size at 1–2% of NAV. Rationale: asymmetric upside if multiple design wins convert to volume; capped-premium downside limits drawdown to option cost. Timeframe: 12–30 months; target return 30–70% if execution and foundry allocation stay on plan; tail risk is full premium loss if a major customer delays.
• Paired trade to hedge customer concentration: Long AVGO stock (1% NAV) / Short GOOGL or META (0.8% NAV) for 9–12 months. Rationale: protects against an AI capex pause that would hit hyperscalers’ orders more directly than Broadcom’s diversified networking business. Risk/Reward: hedged position reduces portfolio beta to an AI-capex shock; if Broadcom converts design wins, expect relative outperformance of 15–40%; if demand weakens, the short helps cap losses to ~20–30%.
• Volatility play: Buy 6–9 month AVGO puts sized to protect 50–75% of the long-equity exposure before the next major product cycle or earnings print. Rationale: insures against a single-program cancellation or foundry reallocation event that can create large, sudden downside. Cost: insurance premium; Benefit: limits drawdown risk while keeping upside exposure intact.