The Billion-Dollar Gamble: Why OpenAI’s Massive Deals with AMD and Nvidia Carry Systemic Risk
The numbers are staggering. In recent weeks, OpenAI, the creator of ChatGPT, has locked in a web of infrastructure and chip agreements with Nvidia and AMD that collectively could exceed $1 trillion. This massive capital frenzy is thrilling the stock market, validating the multi-trillion-dollar AI dream—but it is also sparking uncomfortable comparisons to the notorious dot-com bubble.
Is this the necessary investment for the next industrial revolution, or are we watching financial engineering artificially inflate the world's most valuable companies?
The AI Infrastructure Arms Race
The driving force behind this unprecedented spending is the search for compute power. To build the next generation of AI (like AGI), you need specialized chips, and lots of them. OpenAI, despite a $500 billion valuation, is essentially cash-poor compared to its ambitions, requiring creative financing to secure its supply.
This need has resulted in two landmark, yet complex, deals:
- The Nvidia Nexus: Nvidia, the reigning chip giant with an estimated 80% market share in AI accelerators, has massive financing and investment deals with OpenAI. This setup is often called circular financing: Nvidia invests money into OpenAI, and OpenAI uses that cash to buy Nvidia’s chips. While efficient for ensuring supply, critics worry this mechanism artificially inflates both Nvidia’s revenue and OpenAI’s valuation.
- The AMD Challenge: In a dramatic move to diversify its supply and inject competition into the chip market, OpenAI struck a multi-billion-dollar deal with Nvidia’s rival, AMD. As part of this agreement, AMD will supply up to 6 gigawatts (GW) of computing power (equivalent to the output of several large power plants) and gave OpenAI a warrant allowing it to acquire up to a 10% ownership stake in AMD. This commitment validates AMD's chips as a genuine alternative to Nvidia.
The Dot-Com Deja Vu: Circular Financing
The circular nature of these deals is where the bubble anxiety is centered.
For seasoned market watchers, this structure carries uncomfortable rhymes with the vendor financing practices of the late 1990s tech boom. Back then, telecom suppliers would finance deals for their over-optimistic customers, essentially lending them money to buy their products, thus boosting their own short-term revenues while hiding the lack of true, sustainable profitability.
Today, critics point to the fact that OpenAI, the world’s most valuable unlisted company, is still planning to spend trillions on infrastructure while posting significant operating losses. The central risk is this: if the expected demand for AI services slows down, or if the projected profitability is still years away, this interconnected web of deals could unwind quickly, potentially leading to a sharp market correction.
The Next Great Hurdle: Energy Scarcity and Deployment
Beyond the financial structure, the sheer scale of the power commitment reveals the next major challenge for AI infrastructure: energy.
The promise of 6 GW of AMD chips and a separate 10 GW deployment with Nvidia means OpenAI is planning for computing infrastructure that will consume power equivalent to small nations. This massive demand requires dedicated power plants, substantial grid infrastructure investments, and raises serious ESG (Environmental, Social, and Governance) concerns about sustainability.
As one tech investor cautioned, these billion-dollar agreements are currently "purely announcements, not deployments." The real competition is now shifting:
- Can AMD deliver? It must prove its chips can compete on performance, power efficiency, and scale against Nvidia’s dominant systems.
- Can anyone power it? The logistical and energy hurdles required to bring these gigawatts online are arguably larger than the chip production itself.
The future of the global economy may indeed belong to the company that achieves artificial general intelligence (AGI). But the current frenzy of circular financing and staggering power commitments suggests that the race for AGI may first have to survive an infrastructure bubble—where massive capital expenditure is based on a future that is not yet profitable, demanding careful scrutiny from both investors and regulators.