Nvidia Has 74% of Its Portfolio Invested in 2 Artificial Intelligence (AI) Stocks

Nvidia has 74% of its disclosed investment portfolio concentrated in two AI infrastructure names: 61% in Intel and 13% in CoreWeave. The article is bearish on Intel, citing market share losses, a 150x adjusted P/E, and no external customers yet for its next-generation 14A node, while it is more constructive on CoreWeave due to 73% projected annual revenue growth through 2029 and a 9.6x price-to-sales ratio. Overall, this is commentary-driven analysis rather than a new company announcement, so the likely market impact is limited.

Analysis

The key takeaway is not that Nvidia owns these names, but that it is effectively signaling where scarcity economics still exist in AI infrastructure. CoreWeave looks like the cleaner expression of that thesis: the business is still in an early operating-leverage phase, and if utilization stays tight, incremental revenue should fall through faster than consensus expects. The market is still treating this like a high-growth cloud story, but the more important variable is whether it can sustain pricing power as more capacity comes online.

Intel is the opposite setup: the stock is being priced as if multiple strategic bets can compound at once, but the probability-weighted path still looks poor because the foundry effort competes against better-capitalized, higher-credibility alternatives. A partnership with Nvidia helps sentiment, but it does not solve the core issue that execution lag in manufacturing becomes a structural tax on every product cycle. The second-order winner from Intel’s weakness is not just AMD; it is also ARM-enabled ecosystems that steadily erode Intel’s attach rate in both PCs and servers.

The contrarian miss is that the Nvidia relationship may be overinterpreted as endorsement of long-term winners rather than tactical supplier diversification. Nvidia benefits from having multiple infrastructure options, so its portfolio is more a hedge against bottlenecks than a clean stock-picking signal. That makes the trade asymmetry sharper: CoreWeave can rerate on evidence of margin expansion over the next 2-3 quarters, while Intel likely needs a multi-year proof point that is not yet visible.

Near term, the main risk to CoreWeave is financing pressure, not demand. If rates stay elevated and debt service continues to outpace revenue growth, the equity can de-rate quickly even with strong top-line momentum. For Intel, the catalyst would have to be a meaningful external foundry win or a sharper-than-expected AI attach rate; absent that, the stock can underperform for months even if headline earnings growth looks solid.

Key Decisions for Investors

• Long CRWV vs. short INTC as a relative-value pair for the next 3-6 months; thesis is that AI demand monetization is clearer in CoreWeave while Intel remains a capital intensity and execution trap.
• Add CRWV only on pullbacks or after confirmation of operating margin improvement; target a 6-12 month hold with upside if margins approach management’s 25%+ range, but cut if interest expense continues to outrun revenue growth for another quarter.
• Avoid chasing INTC on partnership headlines; use any strength to reduce exposure or initiate a tactical short against a semiconductor basket if the stock re-rates above 150x earnings without external foundry traction.
• Overweight ARM/AMD relative to INTC for the next 6-18 months; both benefit from share shift in PCs and servers without the same balance-sheet and foundry drag.
• For a defined-risk expression, buy medium-dated CRWV calls financed by INTC short exposure; the convexity favors CoreWeave if AI infrastructure spending stays hot, while Intel’s downside remains tied to slow-moving fundamental erosion.