Nvidia's recent $100 billion investment into OpenAI to support data center expansion has intensified scrutiny of its financial practices. Analysts are now closely examining the chipmaker's strategy of investing in its own customers, a practice that raises questions about the true scale of demand in the booming artificial intelligence market.
These arrangements, often described as "circular financing," involve Nvidia providing capital to companies that then use the funds to purchase Nvidia's products. While not uncommon, the scale of these deals is prompting comparisons to previous technology market bubbles and fueling concerns about inflated valuations across the AI sector.
Key Takeaways
- Nvidia announced a $100 billion investment in OpenAI, adding to a series of large financial commitments to its own customers.
- This practice, known as circular financing, involves investing in companies that subsequently buy Nvidia's products, potentially inflating revenue figures.
- Key examples include investments in data center operators CoreWeave and Lambda, which are also major purchasers of Nvidia's GPUs.
- Analysts are drawing parallels to the dot-com bubble, where similar vendor financing practices preceded a major market correction.
- The complex web of investments makes it difficult to assess the organic demand for AI hardware versus demand subsidized by Nvidia itself.
A Pattern of Strategic Customer Financing
Nvidia's investment strategy extends beyond the high-profile OpenAI deal. The company has established a complex network of financial relationships with key players in the AI infrastructure space, effectively seeding the market for its own advanced processors.
This approach involves direct equity stakes, purchasing cloud services from customers, and providing financial backing that helps startups secure favorable lending terms. While these moves help accelerate the build-out of AI infrastructure, they also create a cycle where Nvidia's capital flows out as an investment and returns as revenue.
The CoreWeave and OpenAI Connection
One of the most prominent examples is Nvidia's relationship with CoreWeave, a specialized cloud provider that supplies data center capacity to companies like OpenAI. Nvidia holds a significant stake in CoreWeave, valued at approximately $3 billion, which represents about 7% of the company.
The financial ties are multifaceted. CoreWeave has purchased at least 250,000 Nvidia GPUs, a transaction valued at around $7.5 billion. In return, Nvidia has committed to purchasing cloud capacity from CoreWeave. This includes an initial agreement worth $1.3 billion and a subsequent deal for $6.3 billion to buy any capacity CoreWeave cannot sell to other clients.
This effectively means that a substantial portion of the money Nvidia has invested in or committed to CoreWeave has returned to Nvidia through chip sales.
The Power of a Co-Signer
According to Jay Goldberg, an analyst with Seaport Global Securities, Nvidia's backing acts like a parental co-signer on a mortgage. It gives lenders confidence, allowing startups like CoreWeave and OpenAI to borrow money for expensive data center projects at interest rates between 6% and 9%, comparable to established giants like Microsoft. Without this backing, they might face rates as high as 15%.
Expanding the Model with Other Partners
Nvidia has replicated this model with other emerging cloud companies. The company reportedly agreed to spend $1.3 billion over four years to rent its own AI chips from Lambda, another data center operator. A separate deal for $200 million was also arranged for additional chip rentals.
These arrangements are viewed by some market observers as evidence of market froth, especially since Lambda reportedly financed its purchase of the GPUs with loans collateralized by the chips themselves.
Nvidia's investment portfolio also includes multimillion-dollar stakes in other public companies that are either customers or operate in adjacent technology sectors, including Arm, Applied Digital, and Recursion Pharmaceuticals. The company also recently acquired a 4% stake in rival chipmaker Intel for $5 billion.
Global Investments and Revenue Projections
The chipmaker's investment strategy has a global reach. In the United Kingdom, Nvidia has pledged £2 billion ($2.7 billion) for AI startups. A significant portion, at least £500 million, is designated for Nscale, a UK-based data center operator that will likely use the funds to purchase Nvidia GPUs.
In 2024, Nvidia's corporate venture arm, NVentures, invested approximately $1 billion in AI startups worldwide. While the exact amount that flows back as revenue is difficult to track, the financial incentive is clear.
A High-Yield Investment Cycle
According to an estimate from Wall Street research firm NewStreet Research, the financial loop is highly profitable for Nvidia. The firm projected that for every $10 billion Nvidia invests in a partner like OpenAI, it could generate approximately $35 billion in GPU sales or lease payments. This potential return would be equivalent to about 27% of Nvidia's entire revenue in its last fiscal year.
This cycle has prompted analysts to voice concerns. Stacy Rasgon of Bernstein Research noted that the OpenAI deal would "clearly fuel ‘circular’ concerns," highlighting the growing unease among financial experts.
Echoes of the Dot-Com Era
For many veteran market watchers, Nvidia's current strategy is reminiscent of the practices seen during the dot-com bubble of the late 1990s and early 2000s. During that period, telecom equipment manufacturers like Cisco, Nortel, and Lucent provided extensive financing to their customers to fuel purchases of their hardware.
"The deals had a whiff of circular financing and were emblematic of ‘bubble-like behavior,'" said Jay Goldberg of Seaport Global Securities, commenting on the perceived similarities to past market excesses.
Before the dot-com bubble burst in 2001, financing extended by major equipment makers to their customers exceeded 10% of their annual revenues. When many of these customers went bankrupt, the equipment manufacturers were left with significant bad debt, which amplified their losses. The networking equipment sector lost over 90% of its value in the subsequent decade.
Distinguishing From Outright Roundtripping
While Nvidia's actions are raising concerns, they are distinct from the more direct "revenue roundtripping" that occurred during the dot-com era. This involved companies like Global Crossing engaging in sham transactions, paying a partner for services in exchange for an equipment purchase of the exact same value, purely to meet quarterly revenue targets.
Nvidia's strategy is more complex, but the underlying concern remains the same: is the demand for AI hardware organic, or is it being artificially stimulated by the primary supplier? By leasing GPUs to OpenAI, Nvidia also takes on the risk of depreciation for these high-cost assets, a burden that would otherwise fall on its customer.
As valuations for AI companies continue to climb, the line between strategic market seeding and unsustainable financial engineering is becoming a central question for investors and the technology industry as a whole.