nVidya, in important ways, is nothing like Enron – the Houston energy giant that imploded through multibillion-dollar accounting fraud in 2001. Nor is it similar to companies like Lucent or WorldCom that went bust during the dotcom bubble.
But the fact that it needs to repeat this to its investors is less than ideal.
Now worth more than $4tn (£3tn), Nvidia makes the specialized technology that powers the world’s AI boom: the silicon chips and software packages that train and host systems like ChatGPT. Its products fill datacenters from Norway to New Jersey.
This year has been an extraordinary year for the company: It has made deals worth at least $125 billion, ranging from a $5 billion investment in Intel – to facilitate its access to the PC market – to a $100 billion investment in OpenAI, the startup behind ChatGPT.
But even as those deals have boosted stock prices and paved the way for Chief Executive Jensen Huang’s energetic world tour, doubts have emerged about how Nvidia does business, especially as it has become increasingly central to the health of the global economy.
These concerns stem from the circular nature of many of its deals. These arrangements are similar to vendor financing: Nvidia lends money to customers so they can buy its products.
The biggest of these is its deal with OpenAI, in which Nvidia will invest $10 billion in the company every year for the next 10 years – most of which will go towards purchasing Nvidia’s chips. Second, there is its arrangement with CoreWave, a company that provides on-demand computing capacity to large AI firms, essentially leasing Nvidia’s chips.
The circularity of these deals has been compared to Lucent Technologies, a telecom company that aggressively lent money to its customers, only to overextend itself and entangle itself in the early 2000s. Nvidia has aggressively denied suggestions of any parity, saying in a recently leaked memo that it “does not rely on vendor financing arrangements to drive revenue”.
James Anderson, a well-known tech investor, describes himself as a “huge fan” of Nvidia, but this year he said the OpenAI deal presents “more reason to be concerned than before.”
He said: “I must say that the term ‘vendor financing’ does not bring up a good reflection for someone of my age. It is not exactly the same as many telecommunications suppliers were in 1999-2000, but there is some rhyme to it. I don’t think from that point of view it makes me feel completely comfortable.”
Other high-profile recent deals include tech firm Oracle spending $300 billion on datacenters for OpenAI in the U.S. — with ChatGPT developer Paying roughly the same amount to use those datacenters. In October, OpenAI and chip maker AMD signed a multibillion-dollar chip deal that also gave OpenAI an option to buy a stake in rival Nvidia.
There has also been a deal with CoreWeave, where OpenAI is receiving $350m in CoreWeave stock, along with a commitment to buy $22 billion of data center capacity from the cloud provider. Asked this month about circularity in the AI industry, CoreWave’s chief executive, Michael Intrator, said: “Companies are trying to address violent shifts in supply and demand. You do that by working together.”
All these moves are part of OpenAI’s $1.4tn bet on computing capacity to build and operate models that, it argues, will transform economies – and make back that spend. OpenAI argues that, while the Nvidia and AMD deals have an investment component, it only begins after the chips have been purchased and deployed, while the investments themselves create incentives aligned to building large-scale AI infrastructure.
Nvidia has also used structures called special purpose vehicle (SPV) in financing deals. The most famous example is the SPV associated with Elon Musk’s xAI: an entity in which Nvidia invested $2 billion, money that will be used to buy Nvidia’s chips.
This drew comparisons with Enron, which used SPVs to keep debt and toxic assets off its balance sheet, convincing investors and creditors that it was stable while hiding growing liabilities.
Nvidia has also strongly denied that it is like Enron: in the same leaked memo where it discussed Lucent, it said that its reporting was “complete and transparent” and that “unlike Enron” it “does not use special-purpose entities to hide debt and inflate revenues”.
Journalist Ed Zitron, a well-known skeptic of the AI boom, agrees that Nvidia Not like any other companyHe says that unlike Lucent, it is not taking on a lot of debt to finance its circular deals, and most of the clients it is supporting are not as obviously risky as Lucent’s dotcom bubble partners, And it’s not like Enron, Zitron argues, because it is quite transparent about its own complex, off-balance sheet deals,
So what could be the need for comparison? “Nvidia isn’t hiding debt, but it is heavily dependent on vendor-funded demand, which creates risks if AI development slows,” says Charlie Dai, an analyst at research firm Forrester. “The concern is about stability, not legality.”
Essentially, whether Nvidia is able to stick the landing depends on whether AI really takes off, generating billions for its corporate users and keeping companies like OpenAI, Anthropic and CoreWave – Nvidia’s customers – firmly in the black, and enabling them to buy their systems. That possibility itself is a matter of debate. If that doesn’t happen, Dai says, Nvidia “could face write-downs on the equity stake and unpaid receivables”: meaning, it could lose a lot of money and its share price could decline.
When contacted for comment, an Nvidia spokesperson referred the Guardian to comments made to investors in early December by its chief financial officer, Colette Cress. Kress said he doesn’t see an AI bubble, but instead points to a trillion-dollar business for Nvidia over the next decade.
In particular, Kress argued that Nvidia’s recent – large-scale deals are just the beginning for the company, and the real money will be earned in the coming years, primarily through replacing almost all chips in existing datacenters with its products.
There’s another complication, which is that the health of Nvidia – and therefore the health of the entire global economy – also depends on whether AI works in time for Nvidia and its customers to repay the debt from their massive datacenter buildouts and significant capital expenditures.
Add to this one final category of concern: recent mega-multibillion-dollar deals with countries like South Korea and Saudi Arabia, the terms of which are opaque. In October, Nvidia said it would supply 260,000 of its Blackwell chips to the South Korean government and South Korean companies. value of this deal Not disclosed, but estimated to be in the billions.
Similarly with Saudi Arabia also. Its state-owned AI startup, Human, has committed to deploying 600,000 Nvidia chips: when that deployment will include actual purchases, and at what price, is again unknown. Nvidia has several other such strategic partnerships – with Italy, with French AI champion Mistral and with Deutsche Telekom, for example – all involving thousands of chips and undisclosed sums.
Payments are likely to be made by governments. There is nothing circular about the sovereign partnership with Germany. But the deals mean more – significantly larger – uncertainties inherent in a tense web of commitments that require massive capital outlays, and rely on ambitious assumptions about an economy undergoing a revolution in the next years.
“They concentrate risk on a few big clients,” says Dai. “If execution is delayed, Nvidia’s revenue recognition and cash flows could be impacted.”
