Prediction: 3 Stocks That Will Benefit More From the AI Boom Than Nvidia by 2028

Broadcom projects more than $100 billion in AI chip revenue in fiscal 2027 — roughly 5x its fiscal 2025 AI revenue — driven by custom ASIC work for hyperscalers and Tomahawk Ethernet networking exposure. AMD has secured commitments from OpenAI and Meta to buy 6 GW of GPUs each (with warrants up to 10% ownership contingent on deliveries/price) and is positioned to benefit from rising agentic-AI demand for high-performance data-center CPUs. Micron, trading under 4.5x forward fiscal 2027 P/E, reported strong fiscal Q2 results and signed its first five-year strategic customer commitment, indicating less cyclical DRAM demand as cloud players lock long-term supply.

Analysis

Broadcom’s architecture- and-IP-heavy model creates disproportionate operating leverage as customers move from brokerage-style procurement to turnkey ASIC programs; the second-order winners are EDA/IP licensors, advanced packaging and OSAT suppliers, and test houses that capture incremental margin as design wins scale. That pathway also shifts capital intensity away from hyperscalers toward foundries and packagers, concentrating execution risk in wafer/packaging throughput rather than in compute-software stacks.

Micron’s move toward multiyear, HBM-like commitments materially reduces revenue variance and turns calendar-year memory cycles into multi-year capacity utilization stories; the immediate read-through is less volatility in FCF and more predictable capex recovery for memory OEMs, but persistent upside hinges on sustained AI unit growth per rack rather than a one-time refresh. Conversely, foundry/packaging bottlenecks or an HBM architecture pivot (e.g., models that reduce memory bandwidth per FLOP) would crush realized pricing and reintroduce cyclical downside within 12–24 months.

AMD benefits from a subtle but powerful demand shift: agentic architectures raise CPU coordination work and I/O orchestration value, which favors incumbents with strong system-level portfolios. The main tail risks across these names are customer-concentration governance (single large hyperscaler program slips), geopolitical export controls on advanced nodes/HBM, and any material easing in model bandwidth intensity — each capable of reversing performance in a single quarter.

From a positioning lens, prefer durable, cash-generative ways to own ASIC/IP exposure and HBM demand rather than long-only high-multiple GPU beta. Timeframe for the structural re-rating is 12–36 months; expect most alpha to accrue when design wins convert to taped-out silicon and show repeatable order streams rather than one-off buys.

Key Decisions for Investors

• Long AVGO (size 0.8% AUM, horizon 12–24 months): initiate with LEAP call spread (buy 12–18 month 0.35–0.45 delta calls, finance with higher strike 0.15–0.25 delta calls) to capture asymmetric upside from scaling ASIC/IP monetization while capping premium. Target 2:1 upside/downside; stop-loss at 30% realized premium loss or 20% adverse move in core networking revs.
• Paired trade — Long AMD / Short NVDA (net size 1.0% AUM, horizon 12–24 months): overweight AMD equity (or 9–12 month calls) and underweight NVDA via short-dated call overwrites or small outright short (if allowed) to express CPU + software coordination upside vs compressed relative multiple on dominant GPU leader. Expect neutral-to-positive carry; tail risk if GPU demand resurges broadly within 3–6 months.
• Long MU (size 0.7% AUM, horizon 9–18 months): buy stock and sell 1–3 month slightly OTM puts to monetize elevated option vol, scaling into confirmed multi-year supply contracts. Risk/reward: asymmetric payoff if HBM contracts stick (potential >100% upside over 12–24 months); downside 40% if cyclical memory glut returns—trim at 50% of position on failure to convert verbal commitments into POs in two quarters.
• Tactical options overlay (size 0.3% AUM): buy AVGO calls funded by selling short-dated NVDA calls to express the structural IP/networking rerating vs. GPU multiple compression. Keep duration mismatch (longer AVGO, shorter NVDA) to exploit mean reversion of volatility and event-driven dispersion; cap collateral and mark-to-market weekly.