Nvidia Becomes Major Shareholder of SpaceX and Intel, Reviving Debate on Circular Risk in AI
13F filing shows Nvidia with $21 billion in SpaceX and $30 billion in Intel, two anchor clients of the company itself. The pattern echoes Cisco's vendor financing model from 1999.
The Numbers from the 13F
On Thursday, August 14, Nvidia released a 13F filing indicating a position of $20.98 billion in SpaceX and $29.99 billion in Intel at the end of the second quarter. The two combined positions account for 80.3% of the $63.44 billion portfolio of the chip maker in listed stocks and private equity. This marks the first time a semiconductor company has emerged as a significant shareholder in both an anchor client of edge inference chips (SpaceX/Starlink) and a competing client foundry (Intel) simultaneously.
The position in Intel shifted from a closed private placement in December 2025 at $23.28 per share to 214,776,632 shares issued, worth $5 billion at entry. At the closing price of the second quarter, this position is valued at $29.99 billion, representing an accounting profit of about $25 billion in less than seven months. The stake in SpaceX arose from the absorption of xAI, in which Nvidia had previously invested $2 billion, by Elon Musk's company earlier this year.
The Echo of 1999
The pattern recalls, but does not replicate, Cisco Systems' model from 1999. In that cycle, Cisco financed the purchase of switches by telecom operators through vendor financing, recognized revenue, and absorbed the liability when the client defaulted. The result was a multi-billion dollar write-down in 2001. The difference today is that Nvidia operates with a gross margin above 70% on datacenter GPUs and uses its own balance sheet, not credit structures, to take a minority position in the client.
This distinction does not eliminate the risk. Exposure is twofold. If SpaceX or Intel underperform, Nvidia loses chip sales revenue and suffers a decline in the market value of the position, increasing operational beta in adverse scenarios. This correlation often requires risk committee approval in other industries before a concentration at such a high level is undertaken.
What Points in the Opposite Direction
Not all analysts view the move as problematic. Sell-side firms with a constructive bias argue that the position in Intel serves as a strategic hedge against a scenario where TSMC and Samsung fail to meet global demand for advanced foundry, and that exposure in SpaceX is a bet on Starlink's dominance as the primary connectivity platform for edge inference, in which Nvidia would benefit as a supplier of embedded chips. The optimistic reading is internally consistent.
Bernstein views the portfolio differently. An 80% concentration in two names, both part of Nvidia's own sales cycle, has characteristics of a sector fund rather than a corporate balance sheet. The same behavior is currently seen in the OpenAI ecosystem, with cross-investments involving Microsoft, SoftBank, and Oracle, reinforcing the criticism that the AI complex increasingly operates as a closed loop.
Global Insight for CIOs and Risk Committees
In the United States, the filing intensifies the debate about systemic concentration in the AI complex. The Federal Reserve has already been citing in its financial stability report the vertical interconnection among a few players in compute, chip, and venture capital. Nvidia's entry as a shareholder in SpaceX and Intel precisely amplifies this characteristic. It is likely to be mentioned in the next semi-annual report.
In Europe, the reading is different. Brussels continues to insist that the lack of a European champion in advanced foundry is an industrial problem. That Nvidia is indirectly financing Intel's recovery through its shareholding may reduce the political appetite in Berlin and Paris for additional subsidies to ASML and STMicroelectronics, changing the economics of the investment plan for fabs on European soil. The IPCEI for microelectronics, still under negotiation, gains a new variable.
In Japan, the effect is indirect yet material. Rapidus, a joint project by Toyota, Sony, Softbank, and others for a domestic 2nm foundry, relies on anchor contracts with Nvidia and AMD to achieve scale. If Nvidia consolidates strategic dependence on Intel through its stake, the hiring timeline for Rapidus may be postponed. This alters the investment schedule for consulting firms contracted for the rollout, especially Accenture Japan and leading local firms specializing in semiconductors.
The Point No One is Making
The less discussed implication is of corporate governance. By becoming a significant shareholder of SpaceX and Intel, Nvidia acquires privileged voice potential in strategic committees of both, which could create conflicts of interest when SpaceX and Intel compete for future contracts from the Department of Defense or equivalent European agencies. The SEC has yet to comment. The first concrete case of conflict will determine whether the circular investment model can withstand regulatory scrutiny or if it needs to be dismantled before the next quarterly close.