The firms at the core of the AI boom are spending billions of dollars on one another, and experts claim that the growing interdependence is raising the possibility of an AI bubble.
As part of a cooperation, Nvidia announced in late September that it would invest up to $100 billion in OpenAI (OPAI.PVT). This would allow the ChatGPT manufacturer to utilize Nvidia’s processors to train and run its next generation of models.
This is just one of a slew of transactions involving a tight network of Big Tech participants that have been made public in recent months. There’s also Nvidia’s $6.3 billion transaction with AI data center provider CoreWeave, a client in which the chipmaker has 7% interest. Nvidia’s rumored $2 billion investment is tied to its client xAI. Then there are OpenAI’s own partnerships with Oracle, CoreWeave, and chipmaker Advanced Micro Devices.
According to experts on Wall Street, the partnerships underline a developing trend: AI infrastructure providers, such as Nvidia, are investing in their clients, who subsequently purchase more products from the infrastructure providers. In other instances, infrastructure clients such as OpenAI are making investments in their suppliers.
Analysts polled by Yahoo Finance expressed two significant worries about the cyclical dynamic witnessed in the current surge of AI investments. For one, the structure of the transactions may give the impression that there is more demand for AI than there actually is. It also strengthens the relationship between the values of Big Tech companies — especially since their individual stocks have skyrocketed on news of such transactions — and intertwined their destinies, such that a hit to any one business would be bad news for the entire ecosystem, according to the experts.
According to Kim Forrest, chief investment officer at Bokeh Capital Partners and a tech expert, the most recent developments are quite concerning. Vendors [of AI infrastructure] have received a lot of money, so they’re simply putting money back into their customers, which may be badly spent.
According to Cornell professor Karan Girotra, examples of suppliers and consumers financially supporting one another weaken the overall system’s “resilience”: “If something goes wrong, then the effect flows through the system rather than [being] isolated.”
Legendary short seller Jim Chanos, famed for forecasting the fall of Enron during the dot-com crisis, also chimed in, saying in a post on X last week, “[Don’t] you think it’s a bit odd that when the narrative is ‘demand for compute is infinite’, the sellers keep subsidizing the buyers?”
The dot-com boom of the late 1990s and early 2000s provided the most glaring illustrations of why such circular investments may be hazardous. Internet service providers (ISPs) raced into the market to provide networks and internet access as the internet grew, but many soon found themselves strapped for cash.
In actions comparable to the recent wave of AI acquisitions, equipment suppliers — routers, switches, fiber-optic cables, and other hardware required to make consumer internet a mass offering – invested in ISPs, their clients, by providing loans and acquiring equity holdings in those businesses. The ISPs might then utilize the loans and equity financing to purchase routers or cables from the equipment manufacturers – a process known as vendor financing.
On paper, the transactions were huge and company was thriving. Equipment manufacturers including Cisco Systems (CSCO), Nortel Networks, and Lucent gave billions of dollars in loans to telecom carriers and internet providers between 1999 and 2001.
Dozens of internet service providers failed when funding dried up because they had such weak underlying financials while being substantially backed by the equipment firms. The equipment sellers had to write off their debt because they had no one to repay them.
During the dot-com bubble collapse and the industry’s downward spiral, suppliers’ poor investments in their own clients exacerbated the effects of their crashes. Bloomberg data shows that the tech-heavy Nasdaq Composite dropped more than 70% between March 2000 and the end of 2002, resulting in a loss of over $3 trillion.
The similarities are not precisely the same. For one, major tech firms now have superior profit margins and have primarily supported their AI-related capital expenditures with strong internal cash flows rather than loans — but analysts believe this may alter as businesses mature. Oracle, a key player in the AI revolution, raised $18 billion in debt late last month.
The system is excessively dependent on the success of OpenAI in particular, according to critics today, because of the complex web of AI investments. The firm that created ChatGPT has not yet made a profit, and analysts are concerned about what may occur if its revenue projections are not met.
In a note dated October 6, Bernstein analyst Stacy Rasgon stated that “[OpenAI CEO Sam Altman] has the power to crash the global economy for ten years or take us all to the promised land, and right now we cannot determine which is in the cards.”
And according to Gil Luria, an analyst at DA Davidson, some of the most recent transactions are especially worrisome since, despite getting funding from Nvidia, AI firms like OpenAI and CoreWeave have either taken on more debt or declared plans to do so, according to Yahoo Finance.
He said that they were raising debt with that funds. The really harmful conduct is the levering up.
Cory Johnson, chief market strategist at Epistrophy Capital Research, added that these kinds of agreements are an indication of a dysfunctional ecosystem: If a consumer needs to borrow money to purchase your goods, they aren’t a good customer.
Circular investments have indeed been around since the beginning of the AI industry. Microsoft (MSFT) contributed $19 billion in contributions to OpenAI between 2019 and the present, while Amazon (AMZN) made two separate investments totaling $8 billion in 2024 to fund AI start-up Anthropic.
Some Wall Street analysts think that such AI market collaborations are beneficial because they enable the necessary cash to fund the AI infrastructure buildout to be deployed more quickly, thus expediting the route to a return on Big Tech’s large expenditures.
He went on, “I don’t think we’re anywhere near bubble territory.”
In response to a recent podcast question concerning assertions that transactions such as Nvidia’s investment in OpenAI, which spends billions as an Nvidia customer, resemble the dotcom bubble’s flawed transactions, Nvidia CEO Jensen Huang maintained that OpenAI’s earnings and any investments the AI company receives are unrelated.
The next multitrillion-dollar hyperscale business is probably going to be [OpenAI], and who wouldn’t want to invest in that? “Heng said.” “My only regret is that we were so impoverished that we didn’t invest when they invited us to do so early on, even though I should have given them all of my money.”
