75% of Nvidia's $63 Billion Investment Portfolio Has Been Put to Work in Just 2 Game-Changing AI Stocks

Nvidia’s AUM rose to $63.4B (+3.4x in one quarter) as its top holdings—Intel at ~$30B (44.23% of invested assets) and SpaceX at ~$21B (30.94%)—grew after prior investments tied to AI data-center buildouts. Nvidia agreed to buy $5B of Intel stock at $23.28/share, and its stake was worth nearly $30B by mid-2026, while its $10B investment in xAI translated into ~122.8M SpaceX shares worth close to $21B after SpaceX’s IPO. The article is constructive on Intel’s turnaround funding but cautions that SpaceX’s ongoing losses and heavy capital burn make that position less secure.

Analysis

The important read-through is that NVDA is behaving less like a pure operating company and more like a platform allocator: it can subsidize ecosystem partners with equity and still preserve balance-sheet flexibility. That tends to reinforce its moat, but the market should be careful not to capitalize unrealized gains as if they were recurring earnings; those marks are sentiment fuel, not cash flow. Over the next 6-18 months, the strategic value is that NVDA can “own” adjacent layers of the stack and make switching costs higher for customers and suppliers alike.

INTC is the cleaner second-order beneficiary because the value of the relationship is not the headline equity stake, but the lower perceived terminal risk. If that relationship translates into design wins or better financing terms, it can lift Intel’s multiple even before fundamentals improve. The flip side is that this only works if execution starts to show up in margins and share gains; otherwise, the partnership becomes a narrative crutch and the stock reverts back to being a cycle/turnaround multiple.

SPCX is the weak link. High capex and negative free cash flow make the valuation highly dependent on access to cheap capital and continued investor appetite; that is a fragile setup if the next liquidity event widens supply or if burn remains elevated. The contrarian miss is that the market may be overestimating how much “strategic halo” matters versus actual monetization. If the private-market enthusiasm fades, the mark-to-market on NVDA’s portfolio could compress quickly without any change to NVDA’s core business.

Key Decisions for Investors

• Long INTC vs short AMD or SOXX for a 1-3 month window: express the view that strategic validation and lower financing risk can expand Intel’s multiple faster than fundamentals deteriorate. Invalidate if Intel’s next earnings fail to show margin stabilization or if AI/server design-win commentary weakens.
• Avoid chasing NVDA on the back of portfolio mark-ups; treat this as a non-operating headline and wait for a pullback if you want exposure. The market is likely over-capitalizing illiquid gains relative to the impact on NVDA’s forward earnings.
• Short SPCX on strength or buy puts if liquidity allows, with a 3-6 month horizon: the setup is vulnerable to continued cash burn, secondary supply, and multiple compression once the IPO novelty fades. Cover if burn meaningfully narrows or if management shows a credible path to self-funding.
• If you already own semis, consider trimming correlation risk by rotating part of the position from high-beta AI suppliers into INTC as a relative-value hedge. The thesis is not that Intel is fixed, but that downside may be better supported than the market implies.