Nvidia's Investment Web Tightens: From Chip Supplier to AI's Landlord, Lender, and Anchor Tenant

Nvidia's $63.4B portfolio and reduced Ohio guarantee reveal a calculated strategy to secure AI demand while managing risk.

The semiconductor giant has quietly transformed into something far more complex than a hardware vendor. Over recent weeks, a pattern has emerged that cuts to the heart of how Nvidia intends to shape — and secure — the artificial intelligence boom: it is increasingly writing checks to the very companies that will buy its chips.

That strategy now spans a sprawling equity portfolio, a series of strategic stakes in power providers and infrastructure firms, and a recalibrated approach to financial guarantees that suggests even the most enthusiastic backer of AI knows where to draw the line.

A Portfolio That Speaks Volumes

The clearest window into this shift came via a 13F filing covering the period to June 30, which revealed a total investment portfolio of $63.4 billion — up sharply from $18.4 billion previously. Two positions dominate: roughly $30 billion in Intel and around $21 billion in SpaceX, together accounting for more than three-quarters of the total holdings.

The Intel stake has grown from a $5 billion investment to its current size within twelve months. The SpaceX position, meanwhile, traces its origins to an investment in xAI that was converted into shares of the space company following a merger. That holding has since slipped to roughly $17 billion in value, reflecting a decline in the share price after the June IPO.

Elon Musk has confirmed that SpaceX plans to build exclusively on Nvidia's forthcoming Vera Rubin architecture — a detail that underscores the real purpose of these holdings. The strategic value lies less in balance-sheet gains and more in locking in future demand for Nvidia's silicon.

The Ohio Guarantee: A Trim, Not a Retreat

That logic also helps explain a headline-grabbing decision this week. Nvidia has cut its financial guarantee for the joint data center project with OpenAI in Ohio from $250 billion to under $120 billion, according to the Wall Street Journal. Reuters reported that a revised deal could be signed as early as the weekend.

The reduction applies to the first phase of the project — roughly 5 gigawatts of a campus ultimately envisioned at 10 gigawatts and over $500 billion in total investment. Nvidia remains part of the infrastructure being developed by SB Energy, a SoftBank subsidiary, and committed in September 2025 to supply at least 10 gigawatts of its own systems, with a potential participation of up to $100 billion.

Reading the Ohio cut as a loss of confidence in OpenAI would be a mistake. The move looks far more like a renegotiation of risk allocation than an exit. Nvidia is simultaneously advancing a financing platform with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR — announced August 10 — that could channel more than $500 billion into AI infrastructure. CEO Jensen Huang has said Nvidia would backstop up to 25 percent of that, roughly $125 billion, articulating the company's philosophy in a single phrase: "In AI, compute is revenue."

The takeaway: Nvidia is not offloading risk so much as redistributing it, shifting away from a single massive Ohio guarantee toward a broader structure supported by Wall Street capital.

Smaller Stakes, Same Playbook

The same pattern repeats at a smaller scale. Reuters reported Friday that Nvidia is in talks to invest up to $3 billion in SB Energy, the SoftBank subsidiary tied to the Ohio data center project for OpenAI, alongside roughly $100 billion in credit support for the campus.

Days earlier, news emerged that Nvidia plans to invest up to $3 billion in Lancium, the power provider behind the Stargate data center in Texas — $2 billion immediately for around 20 percent of the company, with an additional $1 billion contingent on conditions.

Nvidia has also taken a stake in Firmus, an AI infrastructure company that reached a valuation of over $10.5 billion after a $2 billion capital raise, with Nvidia joining Coatue Management, Blackstone funds and Jane Street as shareholders.

The through-line is unmistakable: Nvidia is securing demand not merely as a supplier, but as a co-owner of the demand side itself.

Cautious Signals From the Market's Edges

None of this is without friction. The Ohio guarantee was already scaled back in early August from its original $250 billion figure, a recalibration that suggests even Nvidia recognizes the risks embedded in such mega-commitments.

Skeptics have their own ammunition. Investor Steve Eisman warns that the entire AI boom rests on just two customers — OpenAI and Anthropic — which together account for roughly 70 percent of AI-related revenue at Microsoft, Amazon, Alphabet and Oracle. Michael Burry has taken public positions against Nvidia and semiconductor stocks. Goldman Sachs has flagged roughly $1 trillion in unstarted hyperscaler leasing commitments. Insider sales at Nvidia totaling $767.2 million within 90 days add another layer of caution.

Yet the operational fundamentals remain intact. IBM and Together AI signed a multi-year agreement in mid-August worth $240 million to build an AI cluster using Nvidia HGX B300 systems and Spectrum-X Ethernet technology, with the first expansion phase expected to include around 2,000 Blackwell 300 chips. Nvidia's development of the Nemotron 4 model family — its largest model reportedly featuring at least one trillion parameters — signals an expanding footprint in software ecosystems, not just hardware.

What the Market Is Saying

The share price has settled at €194.74, just 3.8 percent below its 52-week high of €202.50 and roughly 16 percent above its 200-day average. The stock also sits a substantial 39 percent above its yearly low. For now, investors appear to be taking the restructuring of Nvidia's financing architecture in stride rather than questioning it.

The real test arrives with the quarterly earnings report expected August 26, for which Nvidia has guided to around $91 billion in revenue.

The dual role Nvidia now occupies — chip supplier and infrastructure investor — carries inherent tension. The strategy secures demand that might otherwise depend on customers' ability to raise capital, a shrewd move in a market where capital intensity has become the limiting factor. But the growing entanglement between manufacturer and buyer could prove a boomerang in a downturn. The Ohio reduction suggests Nvidia itself senses that boundary.

For now, the growth logic prevails. But the dependencies the company is creating for itself merit closer scrutiny than they have received. The risk hasn't disappeared — it has simply been redistributed across a far wider web of counterparties, from Wall Street giants to power providers in Texas and Ohio. Whether that web holds will likely determine whether Nvidia's gamble as AI's banker, landlord and anchor tenant pays off over the coming quarters.