3 AI Stocks That Can Beat Nvidia Over the Next Five Years

The article argues that Sandisk, Nebius, and AMD could outperform Nvidia over the next five years, citing Sandisk's 61% revenue growth, 672% net income growth, and $4.6B Q3 revenue guidance, plus Nebius' $17.4B Microsoft deal and two Meta contracts worth $12B and $15B. AMD is highlighted for 34% fiscal 2025 revenue growth and a long-term target of more than 35% CAGR, including a 60%+ CAGR goal for its data center business. Overall, the piece is bullish on AI infrastructure names and their earnings potential, though it is framed as opinion rather than new company-reported news.

Analysis

The setup is not just “more AI spending”; it is a supply-chain re-rating where the lowest-capital-intensity bottlenecks may capture the best economics. NAND and neocloud capacity are both becoming toll booths on AI deployment, which means the market can keep mispricing them as cyclical hardware while they behave more like constrained infrastructure with multi-quarter backlog visibility. That favors names with tight supply, visible booking conversion, and pricing power, while it pressures buyers of the underlying compute stack to accept structurally higher depreciation and memory/input costs.

The second-order effect is margin squeeze elsewhere in the AI ecosystem. If memory and colocated capacity remain tight, hyperscalers and GPU buyers will have to choose between faster deployment and lower near-term ROI, which could cap enthusiasm for the largest platform names on every rally. AMD benefits most as the credible share-taker: even modest mix gains versus the category leader can translate into outsized valuation upside because the market is already paying up for any believable second source in a supply-constrained market.

The main risk is timing mismatch. SNDK and NBIS both look strongest on a 6-18 month horizon, but if capacity additions accelerate or demand normalizes, the same scarcity premium can compress fast. NBIS also carries execution risk: contracted megawatts are not the same as revenue-generating, power-delivered capacity, so the market may be capitalizing backlog ahead of infrastructure readiness; any slip would hit the equity harder than the operating data suggests.

Consensus still looks underweight the “picks-and-shovels with pricing power” trade and overfocused on GPU leadership. The contrarian read is that the next leg of AI alpha may come from enabling constraints, not from the dominant accelerator name, because incremental dollars are shifting toward whoever can remove bottlenecks fastest. That makes this a relative-value regime rather than a pure beta trade, especially if AI capex keeps expanding but growth rates slow enough that investors reward durability over narrative.

Key Decisions for Investors

• Long SNDK on 3-6 month horizon; use pullbacks to add, targeting continuation of the scarcity premium, but cut if sequential growth decelerates meaningfully for two consecutive quarters.
• Long AMD vs NVDA pair for 6-12 months; thesis is share gains and valuation re-rating from credible secondary-supplier status, with downside protected if GPU demand broadens rather than concentrating.
• Long NBIS only as a basketed position against hyperscaler capex beneficiaries; best entry is after confirmation of power-to-revenue conversion, because execution slippage is the key risk.
• Avoid chasing MSFT/META on this headline as outright longs; treat them as funding legs in a pair against SNDK/NBIS until AI capex monetization proves it can outrun infrastructure inflation.
• If available, buy 6-9 month call spreads on SNDK or AMD to express upside with defined risk, since both names can gap on order-flow and guidance revisions while implied volatility likely stays elevated.