Nvidia's $63 Billion Investment Portfolio Has a New No. 2 Holding

Nvidia’s 13F showed SpaceX as its new No. 2 public holding (122,764,805 shares worth nearly $21B as of late June), but the outlook is “murky” given heavy cash burn. SpaceX reported a 92% jump in Q2 sales alongside large data-center capex, with AI capex rising from $3.32B in the first half of 2025 to $23.55B in the first half of 2026. The article also flags potential share-pressure from SpaceX’s accelerated, staggered post-IPO lockup/unlock schedule and possible need for dilutive financing.

Analysis

The market should treat Nvidia’s venture book as optionality, not as a driver of intrinsic value. The important mechanism is that headline gains on private stakes can inflate the “AI ecosystem” narrative without changing NVDA’s core earnings power; that makes NVDA less interesting as a direct trade on this story than the underlying capital recipients. If anything, a weak mark on the newly public asset can create a temporary sentiment overhang in NVDA, but it is unlikely to alter valuation for more than a few sessions unless investors start questioning the durability of AI capex broadly.

The more tradable second-order effect is on the newly public asset itself: heavy burn, recurring financing needs, and staged float increases are a classic recipe for multiple compression after the IPO pop fades. That dynamic also pressures adjacent AI infrastructure names by reminding the market that a meaningful share of AI demand is still venture-funded rather than self-funding, which can slow the premium assigned to unprofitable growth across the cohort. On a 1-3 month horizon, any disclosed secondary, convert, or strategic sale will matter more than operating headlines; over 6-18 months, dilution and insider supply are the key variables.

The contrarian point is that bulls may be overestimating the permanence of the AI capital cycle while underestimating Musk-related execution dispersion. If management attention and capital are split across multiple large platforms, the market can keep applying a higher discount rate to the less-proven asset even if top-line growth remains strong. That makes the cleaner expression a relative-value short in the public vehicle rather than a naked bearish view on Nvidia, whose core franchise remains insulated unless AI infrastructure spending itself rolls over.

Key Decisions for Investors

• Short SPCX into any post-IPO strength or ahead of the next unlock window; catalyst path is 1-3 months, with dilution/secondary risk the primary downside driver. Cover if the company prints a materially lower cash burn rate or secures non-dilutive financing that extends runway by >12 months.
• Pair trade: long NVDA / short SPCX as an AI-exposure barbell. Use NVDA as the higher-quality balance-sheet and earnings compounder while fading the venture-like equity funding overhang in SPCX; this is the cleaner relative-value expression if AI enthusiasm remains intact.
• Avoid chasing NVDA on this headline alone; treat any weakness from mark-to-market noise as a buyable dip only if core AI capex guidance from hyperscalers stays firm. Falsifier for the bullish core thesis: a downward revision in 2026 capex plans from major cloud buyers.
• Watch TSLA as a secondary loser from the 'attention tax' on Musk, but do not short it on this catalyst alone. Reassess only if Tesla deliveries or margin guidance soften while AI-related cash demands keep rising across the Musk complex.