NVIDIA’s recent $100 billion investment in OpenAI to support its massive data centre expansion has sparked significant unease and debate in the tech and finance sectors. While framed as a strategic partnership, the deal has raised a critical question for investors: How much of the AI boom is driven by genuine demand versus “circular” financing arrangements from NVIDIA itself?
The complex web of “circular” deals
NVIDIA has a history of investing in its own customers, but the OpenAI deal is by far the most prominent example of this practice. This type of vendor financing, while not unique to the industry, creates a murky financial picture. Analysts are concerned that these deals may give the public and investors an inflated perception of the true market demand for AI hardware.
This tangled web of investments is difficult to untangle. For instance, NVIDIA has invested in both OpenAI and CoreWeave, a company that provides data centre capacity to OpenAI. Meanwhile, CoreWeave is also a major buyer of NVIDIA’s GPUs. This creates a chain of money where NVIDIA’s cash flows out as an investment and then flows back in as revenue from its customers.
The risks of self-financing
This practice raises concerns reminiscent of past technology bubbles, particularly the dot-com bust of the early 2000s. In that era, telecom companies engaged in “revenue roundtripping,” where they would lend money to customers to buy their equipment, artificially inflating revenue and demand. When the bubble burst, the companies were left with bad debt and massive losses, leading to a market-wide collapse.
While the current percentage of NVIDIA’s revenue from these circular deals is believed to be small, its dominant position as a trillion-dollar company means even minor financial missteps could have a massive impact on its valuation and the broader economy.
NVIDIA’s dominance and market power
The benefits of a NVIDIA investment extend beyond just cash. A company like OpenAI or CoreWeave with NVIDIA’s backing can secure loans for data centre projects at significantly lower interest rates than they would otherwise be able to get. This acts as a form of co-signing, providing lenders with a sense of security that their loans will be repaid.
NVIDIA has also entered into extensive cloud computing deals, agreeing to rent out its own chips from companies like CoreWeave and Lambda. In one notable instance, Lambda bought NVIDIA chips using borrowed money, then rented the chips back to NVIDIA, using the rental income to pay off the loan. For sceptics, this is a clear signal of speculative, bubble-like behaviour.
Echoes of Past Bubbles
The long-term success of this business model depends on sustained high demand for AI technology. By leasing GPUs to customers, NVIDIA is taking on the risk of asset depreciation and the possibility of being stuck with unwanted inventory if the AI boom cools. This contrasts with a traditional sales model where the customer takes on that risk by buying the chips outright.
As AI valuations continue to climb, analysts are openly questioning whether this circular financing is a sign of a healthy market or a ticking time bomb. While the distance between concern and crisis is still significant, the intricate financial relationships between NVIDIA and its customers are raising serious questions about the true financial stability of the AI industry.
Sources
- Yahoo Finance – NVIDIA’s OpenAI Investment Analysis
- Benzinga – Expert Warnings on NVIDIA’s Investment Strategy
- Fortune – Analysis of AI Investment Patterns
