NVIDIA announced earlier this week that it exists Investing 100 billion dollars In Openai to help finance the huge data center, it added to a growing feeling of discomfort among investors that there is a dangerous financial bubble about artificial intelligence, and that mathematics and profits revenues that support the evaluation of both public and private companies in the sector only.
Although the latest NVIDIA advertisement is largely the biggest examples, artificial intelligence maker has participated in a series of “circular” deals in which you invest or give money to its customers. The financing of the sellers is somewhat extent in many industries, but in this case, circular transactions may give investors an enlarged perception of real demand for artificial intelligence.
In previous technology bubbles, “round” revenues and technology companies that finance their customers exacerbated the damage when these bubbles appeared in the end. Although the share of NVIDIA’s revenue currently drives such funding appears relatively small, the company’s dominance as the most valuable company in the world is in general means that its shares are “price for perfection” and that even simple errors can have a significant impact on its evaluation-on financial markets and perhaps even the wide economy.
It is not easy to answer accurately, the extent of the entire artificial intelligence boom by Nafidia specifically, and is also one of the annoying things on this topic. The company has entered into a number of investment and financing deals, many of which are very small individually so that the company cannot consider “materials” and the report in its financial files, although it may be largely large.
In addition, there are many interlocking episodes of generalization – where NVIDIA has invested in a company, such as Openai, so that it in turn purchases services from a cloud service provider that NVIDIA has also been invested, which then also Buy or rent graphics processing units from NVIDIA – this imposes what money flows anywhere far from the easy.
Interlocking investment networks
Two of the most prominent examples of the NVIDIA network of circular investments are Openai and Coreweave. In addition to the latest investment in Openai, NVIDIA previously participated in a $ 6.6 billion investment tour at the Quantic IQ in October 2024. NVIDIA also invested in Coreweave, which provides the ability of the Data Center to Openai and is also a NVIDIA agent. As of the end of June, NVIDIA has about 7 % of Coreweave, a share of about $ 3 billion currently.
The benefits that companies receive from NVIDIA investing the same money. These companies such as Openai and Coreweave shares these companies allow these companies to access debt financing for data center projects at much lower interest rates, which may be able to reach without this support. Jay Goldberg, an analyst at Seaport Global Securities, compares such deals with someone who asks their parents to be involved in their mortgage. It gives the lenders some assertion that they may already recover their money.
Startup financing data centers often have to borrow money at rates of up to 15 %, compared to 6 % to 9 % that a large company like Microsoft may have to pay. With NVIDIA support, Openai and Coreweave can borrow at rates closer to what microsoft or Google might pay.
NVIDIA has also signed a $ 6.3 billion deal to buy any cloud energy that Coreave cannot sell to others. The chips maker had previously agreed to spend $ 1.3 billion over a period of four years on cloud computing with Coreave. Meanwhile, Coreave has bought at least 250,000 NVIDIA graphics processing units so far – most of which say they are Hepper H100 models, which cost about $ 30,000 each – which means that Coreove has spent about $ 7.5 billion in buying these chips from NVIDIA. Therefore, in essence, all the money that invested NVIDIA in Coreweave has returned to that in the form of revenues.
NVIDIA has concluded similar cloud computing offers with other “Neo-CLOUD” companies. According to a story in the information, NVIDIA agreed this summer to spend $ 1.3 billion over a period of four years to rent about 10,000 of its artificial intelligence chips from Lambda, which such as Coreave runs databases, as well as a separate deal of $ 200 million to rent about 8,000 again over an unspecified period of time.
For those who believe that there is an Amnesty International bubble, the Lambda deal is clear evidence of butter. Those chips Nvidia Lambda rent the time to return to Nvidia? I bought it with borrowed funds guaranteed with the value of the graphics processing units themselves.
Besides its large investments in Openaii and Coreweave, AI Chipmaker owns millions of dollars in many other companies circulating for the public that buy either their graphics processing units or work on relevant chips technology. This includes ARM Design Design Arm, the HD Applied Performance Company, the Nebius Group Cloud, and Recursion Parmaceuticals in the field of biotechnology. (NVIDIA recently bought a 4 % stake in Intel For $ 5 billion. Like ARM, Intel makes chips in some cases that are alternatives to the NVIDIA graphics processing units, but in most cases they are complementary to them.)
Earlier this month, NVIDIA also pledged to invest two billion pounds ($ 2.7 billion) in startups in the United Kingdom, including at least 500 million pounds in NSCale, a UK -based data center, which is supposed to be used some of these funds to buy NVIDIA GPUS to provide the data centers that it adopts. NVIDIA also said that it will invest in a number of British startups, through local investment capital companies, and some of these funds as well, are likely to return to Openai in the form of computing purchases, either directly, or through cloud service providers, who in turn need to buy Nvidia GPU.
In 2024, NVIDIA invested about one billion dollars in the AI startups in the world either directly or through the corporate investment capital arms, according to data from Deal Room and Financial times. This amount increased significantly from what NVIDIA invested in 2022, the year in which the Tuwaidi artificial intelligence boom began with Openai for the first time in ChatGPT.
The amount of this money ends with return directly to NVIDIA in the form of sales again, difficult to determine. Wall Street Research has estimated that for every $ 10 billion of investment in NVIDIA in Openai, GPU purchases will witness $ 35 billion or GPU rents, with an amount equal to about 27 % of its annual revenues in the past fiscal year.
Dotcom echoes era
Certainly, this type of return will make this type of customer financing worth it. But it raises concerns among analysts about a bubble in assessing artificial intelligence. These types of circular deals were a distinctive feature of previous technology bubbles and often returning to the investors.
In this case, the lease arrangements that NVIDIA make with Openai may be proven as part of its latest problem. By renting graphics processing units to Openai, instead of asking them to purchase chips directly, NVIDIA reaps Openai from having to take accounting fees against high -consumption rates on chips, which will eventually help in Openai. But this means that instead NVIDIA will have these consumption costs. Moreover, NVIDIA will also be at risk of being exposed to graphics processing units, no one wants if the demand for Amnesty International’s work does not coincide with Rose Ginsen Huang predictions.
For some market monitors, the latest NVIDIA deals are completely similar to the excesses of previous technological mutations. During the Dot Com bubble at the end of the twenty -first century, communications makers such as Nortel, Lucent and Cisco Money lend to startups and telecommunications companies to buy their equipment. Immediately before the bubble explosion, the CISCO and Nortel financing amount extended to its customers 10 % of the annual revenue, and the amount of funding for the largest of the five equipment makers provided to customers exceeded 123 % of their joint profits.
Ultimately, the amount of cables and switching the fibers that were installed for the request exceeded, and when the bubble and many of these customers exploded, the communications equipment makers carrying regular debts on their public budgets were left. This contributed to losing more value when the bubble exploded than it was, as companies have lost more than 90 % of their value during the next decade.
Worse, companies such as Crossing Global Global Global fibers that participate in direct “round revenues”. These companies reduced deals – often at the end of the quarter to reach Topline’s forecasts – in which they paid money to another services company, then that company agreed to purchase equipment of completely equal value. When the bubble exploded, Global Crossing went bankrupt, and the executives ultimately pushed a major legal settlement related to round revenues.
They are the memories of these types of transactions that have caused at least raising analysts in some circular investments in NVIDIA. Goldberg, global port analyst, said the deals have a whiff of circular financing and was symbolic “for the bubble behavior.”
“The procedure will clearly support” circular concerns “,” wrote in an investor note, “The procedure will clearly support” circular interests “,” Wipenai, in an investor note after NVIDIA’s announcement of its popular investments in Openai. Of course, from anxiety to the crisis, of course, but with the high assessments of the artificial intelligence company, this distance began to close.
https://fortune.com/img-assets/wp-content/uploads/2025/09/GettyImages-2225512564-e1758898853117.jpg?resize=1200,600
Source link