Nvidia disclosed a $21 billion stake in SpaceX and a $30 billion stake in Intel in its Q2 FY2026 regulatory filing, with the SpaceX position traced back to Nvidia's earlier investment in xAI, which SpaceX absorbed in a $1.25 trillion all-stock merger.

The most counterintuitive detail in this filing is that Nvidia didn't decide to become a SpaceX shareholder - it woke up as one. The 122.8 million SpaceX Class A shares trace back to Nvidia's roughly $10 billion contribution to xAI's $20 billion Series E round, closed in January 2026 ^[1]. When SpaceX absorbed xAI in an all-stock deal valued at $1.25 trillion in February 2026, every share of xAI, Nvidia's included, converted directly into SpaceX Class A stock ^[3]. Nvidia's regulatory filing on August 14 simply reported the position as it now stood ^[2]. That distinction matters for how the story gets read: this isn't Nvidia buying influence over a rocket company, it's a chip investment that got swept into an unrelated merger and came out the other side worth billions more.

The disclosure landed squarely inside an ongoing argument about whether Nvidia's equity stakes in its own chip customers amount to healthy strategic alignment or a self-reinforcing loop. Nvidia CEO Jensen Huang has pushed back directly, arguing the underlying demand for Nvidia's products is real and that the capital involved isn't recycled Nvidia revenue ^[4]. Bank of America's Vivek Arya framed Nvidia's separate $500 billion MOU with six financial institutions as a pivot away from the vendor-financing structure that drew circularity criticism in the first place, but Wells Fargo's trading desk stayed unconvinced, noting that in the end Nvidia is still part of the financing ^[4]. Investor Michael Burry has gone further, describing the pattern across Nvidia's stakes and multi-year purchase commitments as a circular financing web worth an estimated $46 billion in direct equity and $879 billion in commitments ^[4].

Strip out the merger mechanics and the raw numbers are striking on their own. Nvidia's roughly $10 billion Series E contribution to xAI converted into 122.8 million SpaceX shares ^[1][5], working out to an effective entry price in the neighborhood of $80 a share. Against SpaceX's June 30 close of $170.86, that implies the position had roughly doubled in value before Nvidia ever filed anything about it ^[6]. That paper gain proved fragile. SpaceX had only completed its IPO on June 12, pricing shares at $135, which put Nvidia's converted stake in the $16 to $17 billion range at the very start of public trading ^[8], and within a day of the August 14 disclosure the market moved again: by August 15, SpaceX shares had fallen to about $140, cutting the SpaceX stake to roughly $17.2 billion, while Intel's position had narrowed from $30 billion to about $22 billion over the same window ^[3][7]. Nvidia is still sitting on a large unrealized gain relative to what it effectively paid for the SpaceX shares, but the swing illustrates how much of this story is a snapshot of a single balance-sheet date rather than a stable, current valuation.

SpaceX and Intel together now account for almost 80% of Nvidia's publicly disclosed stock portfolio ^[3][6], which otherwise consists of smaller positions in companies like Coherent, Nebius, Nokia, and Synopsys. That concentration cuts both ways: it ties the value of Nvidia's investment book directly to the fortunes of two companies it also sells chips to, and it reinforces a lock-in dynamic where SpaceX and Intel are described as exclusive buyers of Nvidia's chips even as Nvidia holds equity in both ^[4]. A stumble at either company doesn't just affect Nvidia's customer relationship, it shows up on Nvidia's own books.

The disclosure landed very differently depending on which audience was reading it. Financial media and market commentators largely treated the numbers as routine strategic news, framing Nvidia's expanding equity book across SpaceX, Intel, and smaller cloud and compute partners as evidence the company is deepening ties with the infrastructure players central to AI's next buildout phase, with some commentary framing Nvidia as poised to become SpaceX's exclusive AI-compute supplier, including for future AI satellites. Independent analysts leaned on a keiretsu framing, the postwar model of companies cross-holding each other's shares, to explain how a supplier owning equity in its own customer can blunt normal arm's-length pricing dynamics, while noting the SpaceX position was inherited through the xAI merger rather than actively sought. Retail investor communities on stock-focused forums took a far more skeptical line, reaching for terms like circular financing, shell game, and even Enron comparisons to describe a chipmaker holding equity in the same companies that buy its chips. Some of that skepticism rested on a misunderstanding - several discussion threads initially treated the SpaceX stake as a fresh purchase rather than merger-inherited equity, before the correction spread. A counter-read gained traction too: that the position functions less as financial engineering and more as a competitive moat, discouraging Elon Musk from building an in-house chip alternative and keeping that business away from Nvidia's rivals, though skeptics of that argument pointed out that a stake under 2% of SpaceX buys little real influence either way. The gap between the two readings is itself telling: the closer an audience sits to daily SpaceX and Nvidia stock discussion, the less willing it is to take the arrangement at face value.