Key Takeaways: Nvidia's $63.44 billion portfolio — nearly half in Intel — now swings earnings on equity stakes as much as GPU sales.
Key Takeaways: Nvidia's $63.44 billion portfolio — nearly half in Intel — now swings earnings on equity stakes as much as GPU sales.

Nvidia's investment portfolio surged 245 percent to $63.44 billion as of June 30, with Intel and SpaceX accounting for 80 percent of holdings, reshaping the chipmaker's earnings profile ahead of its Aug. 26 report.
Morgan Stanley reiterated a "Buy" rating with a $288 price target, citing order-book visibility for fiscal 2027, while Bank of America set a more aggressive $350 target, calling Nvidia the leading name in AI infrastructure.
The Intel position alone is worth $29.99 billion — 47.3 percent of the portfolio — while SpaceX stands at $20.98 billion and CoreWeave at 7.4 percent. The portfolio contributed $15.9 billion in other income last quarter, roughly 30 percent of operating profit of $53.5 billion.
The Aug. 26 earnings report will test whether the market can reconcile Nvidia's sprawling construction — equity stakes, land rights, power agreements, and billion-dollar guarantees — with the clean economics of selling chips. Consensus stands at $91.90 billion in revenue and $2.08 in earnings per share.
The concentration reflects a coherent thesis: AI workloads require not just accelerators but also CPUs, cloud capacity, and specialized data centers. Intel has delivered a dramatic rerating in 2026, with shares up 184 percent year to date after second-quarter revenue of $16.1 billion, up 25 percent year over year — the strongest quarterly growth in more than 15 years. CEO Lip-Bu Tan purchased 105,263 shares at $95 per share in early August, a $10 million transaction that coincided with Intel's $20 billion equity offering at the same price.
SpaceX, which went public June 12 in the largest IPO in history, reported AI revenue of $2.6 billion in the second quarter, up 247.5 percent year over year. But the company remains unprofitable, with AI capex vaulting from $3.32 billion to $23.55 billion in the first half of 2026. Its accelerated and staggered lockup schedule could increase float and weigh on shares.
Nvidia is no longer behaving like a conventional semiconductor vendor. Last Friday, it announced financing platforms alongside Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR to mobilize more than $500 billion in third-party capital for data centers. The company also secured land, power capacity, and building shells with SB Energy at the PORTS-Pike Technology Campus in Pike County, Ohio, and is in talks to invest up to $3 billion in the SoftBank subsidiary.
Nvidia has also revised its support plan for the proposed OpenAI data center: instead of the previously discussed $250 billion guarantee, the company would now initially back less than $120 billion. That reduction reads as calibration — a company stress-testing how much balance-sheet exposure it can absorb without overextending.
Management has said the new "Vera Rubin" chip architecture will ship for the first time in the third quarter of 2026, explicitly framed as a measure to limit transition risks with persistent supply constraints. The stock trades at €194.86, roughly 3.8 percent below its 52-week high of €202.50 set May 14, and about 16 percent above its 200-day moving average. Year to date, shares are up 22 percent.
Analyst targets heading into the next earnings report span a wide range. KeyBanc, Bernstein, and TD Cowen sit between $275 and $330. The dispersion itself is telling: even declared optimists cannot agree on how much Nvidia's equity portfolio and infrastructure bets are worth.
Supporting signals have come from customers as well. In early August, Elon Musk said SpaceX would rely exclusively on Nvidia chips going forward. The expanded strategic partnership with SK Group, announced July 24, covers AI factories and advanced memory chips.
For Nvidia shareholders, the positions offer indirect exposure to some of the most closely watched names in AI. But the concentration also means the portfolio's performance will be heavily influenced by a handful of outcomes. Intel's forward earnings multiple of 80.7 times, versus an average of 21.7 for information technology stocks, leaves little room for error. CoreWeave's customer concentration and unprofitability present similar risks. And SpaceX's valuation assumes flawless execution in a competitive AI computing market.
The bull case holds as long as the Vera Rubin architecture ships on schedule and the Intel and SpaceX positions retain their value. The risk is that the second stand-alone business story begins to overshadow the first.
This article is for informational purposes only and does not constitute investment advice.