The Giant Partnership and the Attempt to Create a Secondary Market for Chips
According to a TechCrunch report, Nvidia has announced this week that financial giants Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR are willing to commit up to $500 billion toward building AI data centers. While this eye-popping figure captured the bulk of public attention, the larger and more significant story lies in Nvidia's concerted efforts to establish a secondary market for its aging graphics processing units (GPUs).
To convince these top-tier financial firms to participate in this massive venture, Nvidia has agreed to guarantee, using its own capital, that the chips serving as collateral in these deals will retain their value. This plan has been quickly described by many market observers as highly unusual, clever, and dangerous all at the same time. Indeed, anxieties in the bond markets were deep enough that Nvidia CEO Jensen Huang felt compelled to take to the social media network X (formerly Twitter) and financial television channels to explain in greater depth how the company's actual risk exposure would be limited. Underneath the complex financial maneuvering designed to fund AI data centers and keep Nvidia's revenue flowing, there is a process that could be particularly intriguing for startups and enterprises alike: Huang wants to ensure that a healthy ecosystem of used AI hardware flourishes, thereby helping to sustain demand for Nvidia's hardware even as it ages.
כיצד עובד מנגנון הערבויות והסיכון ההפוך של אנבידיה
Under the newly introduced mechanism, Nvidia guarantees that if the GPUs used as collateral in these transactions do not retain their value as expected, the company will cover up to 25% of the difference. In practical terms, if a data center owner defaults on a loan and the lender is forced to liquidate and sell off the assets, but the chips cannot command the price recorded on the books, Nvidia will step in and pay its share (up to a quarter of the value gap).
This element of the program carries significant risk for Nvidia, a phenomenon known among financial professionals as "wrong way risk." The danger lies in the fact that Nvidia's financial obligations are poised to grow precisely when general market demand weakens. Such a scenario could lead to a squeeze on the company's revenues at the exact moment its liabilities to pay out these guarantees increase. Nevertheless, the program has been deliberately structured to differ from recent market comparisons to Lucent Technologies. Lucent, a telecommunications equipment provider, rose and spectacularly collapsed during the dot-com bubble after directly lending money to its own customers so they could purchase its products and hardware.
צל העבר של לוסנט וההבדלים במודל המימון
The comparison to Lucent casts a persistent shadow over Nvidia, and CEO Jensen Huang is reported to be well aware of this and even understands that the comparison is not entirely unfair. Nvidia has indeed committed billions of dollars toward entities purchasing its chips, including leading frontier AI labs like OpenAI and Anthropic, new cloud companies (often called "neoclouds") such as CoreWeave (which pioneered the practice of using Nvidia chips as collateral to secure loans), as well as other players like Nebius, Firmus, and Lambda. Furthermore, according to calculations by the Bloomberg news agency, Nvidia worked during this past summer on additional circular deals valued at approximately $750 billion.
"Is this circular financing?" Huang wrote on his account on X, referring to the new setup. "This initiative is designed to address that concern. We are bringing independent, long-term institutional capital into the AI infrastructure market." These comments reflect the fundamental distinction from the Lucent model: unlike Lucent, Nvidia is leveraging external, independent financial institutions to shoulder the vast majority of the capital and risk, while Nvidia itself only agrees to protect a relatively small portion (up to 25%) of its chips' future residual value.
הלחץ הפיננסי על ענקיות הענן וההשוואה ההיסטורית לרכבות
In the event that the current plan succeeds, Nvidia will have secured new funding sources for building AI data centers, coming at a time when several traditional methods are beginning to wear thin. For instance, some of the major cloud hyperscalers have already taken on exceptionally heavy debt loads (such as Oracle), issued new tranches of equity (such as Google), or burned through massive volumes of cash (such as Meta). The situation has become so sensitive and complex that Microsoft CEO Satya Nadella recently recommended the book "1873" during his company's latest earnings call. The book focuses on the complex financial engineering of the railroad era, which ultimately triggered a collapse of the American economy during that period.
The overriding risk is that the current AI boom, where present demand vastly outstrips available capacity, may not continue for much longer. Rather than being in the early stages of long-term growth, what happens if enterprises and consumers begin to moderate and scale back their use of AI technologies? Alternatively, what if new technologies are developed that make existing infrastructure many times more efficient, or perhaps render today's entire AI infrastructure completely obsolete? In such a scenario, much like buggy whip manufacturers facing the rise of the automobile (paraphrased by Danny DeVito in his role as Lawrence Garfield), demand could dry up and the entire structure could collapse.
חזון "מפעלי הבינה המלאכותית" והשוק המשני
Despite these deep-seated concerns, Huang argues that such a collapse will not occur. He presents a vision where artificial intelligence represents a long-term "investable infrastructure." This perspective treats Nvidia's AI servers as "AI factories," which are akin in nature to railroads or airlines rather than rapidly depreciating assets like personal computers (PCs).
"When needs change, the factory can be used by another customer, another cloud or another operator. This broad ecosystem gives NVIDIA compute a deep market of potential users and offtakers, helping protect residual value," Huang promised. In this envisioned future, Nvidia attributes as much importance to its aging architecture as it does to its newest cutting-edge chips. Consequently, startups, enterprises, and even researchers may benefit from access to a wider variety of hardware, where each component is custom-tuned to different AI needs—just as they are currently starting to adopt accessible and affordable open-weight models alongside frontier options. As the undisputed leader of the AI landscape, Nvidia possesses both the market power and the current window of opportunity to make this ambitious vision a reality.