$2.2 Trillion Question: The Circular Finance Threatening the AI Boom
Nvidia’s market capitalization, exceeding $2.2 trillion as of January 2nd, now surpasses the GDP of all nations globally except the United States and China. This single figure encapsulates the extraordinary, and increasingly precarious, ascent of artificial intelligence. While headlines trumpet AI’s potential, a closer examination reveals a complex web of interconnected investments and supply chain commitments – a latticework so intricate that even participants are voicing concerns about its sustainability. Follow the money, and a pattern emerges: a series of circular deals designed to prop up growth, but potentially vulnerable to a swift and destabilizing correction.
The most prominent example is the still-unfinalized $100 billion, ten-year agreement between Nvidia and OpenAI, where the chipmaker would invest in the AI developer while OpenAI commits to purchasing 10 gigawatts of AI infrastructure. Nvidia’s November quarterly report, acknowledging “no assurance that we will enter into definitive agreements,” didn’t immediately impact its stock, but it signaled a growing unease about the foundations of this AI-driven expansion. This isn’t isolated; it’s a symptom of a broader trend where investment isn’t necessarily driving independent value creation, but rather reinforcing existing positions within a closed ecosystem.
This dynamic attracted the attention of Michael Burry, the investor who famously predicted the 2008 housing market crash. In December, Burry revealed short positions in both Nvidia and Palantir, citing Nvidia’s customer financing as reminiscent of Enron’s accounting practices before the dot-com bubble burst. This comparison isn’t merely rhetorical; it points to a core issue: are these investments genuinely based on projected future earnings, or are they designed to artificially inflate demand and justify valuations? The historical parallel suggests the latter is a distinct possibility.
Reporting from gfmag.com informs this analysis.
The circularity extends beyond Nvidia and OpenAI. Microsoft and Nvidia jointly invested $15 billion in Anthropic, which in turn is committed to spending $30 billion on Microsoft’s cloud services and – crucially – Nvidia’s chips. Advanced Micro Devices (AMD) offered OpenAI warrants to purchase its stock, simultaneously securing OpenAI as a customer for its chips. These arrangements aren’t organic market forces at play; they are carefully constructed dependencies. The sheer scale of these commitments – Oracle’s $300 billion partnership with OpenAI to develop AI infrastructure, for example – raises questions about deliverability and the potential for cascading failures.
The financial strain is already visible. Oracle’s stock price plummeted 30% in the third quarter of 2024, fueled by concerns about its ability to fulfill its five-year, $300 billion compute capacity commitment to OpenAI, and OpenAI’s ability to pay for it. This isn’t simply a company-specific issue. As Gregory Blotnick of Columbia Graduate School of Business highlighted, the interconnectedness of these investments means trouble for one player can rapidly spread. A slowdown in Microsoft’s AI monetization, for instance, could reduce Azure spending, impacting Nvidia’s revenue, devaluing CoreWeave (5% owned by Nvidia and a key cloud provider for OpenAI), and ultimately jeopardizing OpenAI’s funding capacity.
The current investment frenzy dwarfs even the dot-com boom. AI accounted for roughly half of all US venture capital commitments in the first seven months of 2025, and a McKinsey & Co. report estimates $5.2 trillion will be needed for data center investments by 2030. However, a critical “accounting mismatch” exists, as noted by Mihir Kshirsagar of Princeton University’s Technology Policy Clinic. AI chips have a lifespan of just one to three years due to rapid obsolescence, yet companies depreciate them over five to six years, artificially inflating reported earnings. This practice, combined with the bundling of chip and data center costs into broader construction projections, obscures the true cost of AI infrastructure.
The fundamental question, as Bryan Routledge of Carnegie Mellon University’s Tepper School of Business frames it, isn’t simply how to fund AI, but whether the investment is worthwhile. While capital is more abundant today than during the late 1990s, the risks remain. The interconnected supply chain – chips, data centers, energy – must be built in concert, and regulatory hurdles add further complexity. Moreover, the rapid pace of technological change introduces uncertainty. OpenAI’s decade-long commitment to Nvidia could become a liability if AI technology evolves in unforeseen directions.
The potential for a downturn is real. If the hype surrounding large language models fades, Nvidia’s order book could collapse, mirroring the fate of overvalued telecom companies in the early 2000s. The creative financing now prevalent could then become a source of danger, exacerbating the adjustment. However, the sheer scale of investment by tech giants like Google, Microsoft, Meta, and Amazon – generating $451 billion in operating cash flow in 2024 – provides a degree of resilience.
What this means for your wallet: watch for a potential divergence between AI hype and actual profitability. The current market prioritizes market share over immediate returns, but that can’t last indefinitely. The key question isn’t whether AI will be transformative, but when – and whether the companies currently leading the charge can navigate the complex financial landscape they’ve created without triggering a broader market correction. Specifically, investors should monitor Oracle’s ability to deliver on its commitments to OpenAI and the overall depreciation rates of AI infrastructure. A significant downward revision in these figures could signal a looming reckoning.
