Earlier this week, NVIDIA reports that it is investing $ 100 billion in Openi to help finance her huge data center, increased investors’ anxiety that there is a dangerous financial bubble around AI, and that income and income based on both state and private companies are just unmatched.
Although the last NVIDIA message is the biggest example, the AI minibus manufacturer is engaged in “circular” transactions in which he invests in his customers or lends money. Sellers’ funding exists to a certain degree in many industries, but in this case, ring transactions can give investors a bloated perception of true AI demand.
In previous technology bubbles, revenue “round” and technology companies that fund their customers increased the damage when those bubbles eventually jumped out. Although Nvidia’s revenue, which is currently due to such funding, seems to be relatively small, the dominance of the company, as the world’s most valuable publicly traded company means that its stock is “priced for perfection” and that even minor mistakes may have a major impact on its evaluation-and perhaps even wavy economies.
How much AI boom has been caught by NVIDIA cash is not easy to answer accurately, which is also one of the alarming things. The company has entered into a number of investment and funding transactions, many of which are too small for the company to consider “material” and report in its financial applications, although they may be significant.
In addition, there are so many interrelated circular rings – where NVIDIA has invested in a company such as Open that in turn buy services from a cloud service provider, which he also invested in NVIDIA, which he then invested and then invested at NVIDIA. Invested in Nvidia. too Buying or renting a GPU from Nvidia is to distinguish between what money is flowing where it is far from easy.
The two brightest examples of NVIDIA online investment network are Openai and Coreweave. In addition to the latest investment in Openai, NVIDIA has previously participated in $ 6.6 billion investment in AI, a fast -growing company in 2024. October At the end of June, Nvidia owned about 7% Coreweave, currently shares worth about $ 3 billion.
The benefits that companies receive from Nvidia’s investment exceeds the cash itself. NVIDIA Property Self -Papers campaigns in companies such as Openai and Coreweave provide these companies access to debt funding for data centers with significantly lower interest rates than they could reach such support. Jay Goldberg, an analyst at Seaport Global Securities, compares such transactions to what is asking for their parents to be signing their mortgage. It gives lenders a little assurance that they can actually recover money.
Beginner financing data centers often required borrowing money for 15% rates compared to 6% – 9%, which may have to be paid to large, established corporations such as Microsoft. NVIDIA was able to borrow at the rates at the Opeenai and Coreweave to the rates that Microsoft or Google can pay at the rates.
NVIDIA has also signed a $ 6.3 billion deal to acquire any cloud capacity that Coreweave cannot sell to others. The public wood manufacturer had previously agreed to spend $ 1.3 billion in four years in a cloud computing with Coreweave. Meanwhile, Coreweave has so far purchased at least 250,000 NVIDIA GPU – most of them say the H100 Hopper models, each cost around $ 30,000, meaning that Coreweave has spent about $ 7.5 billion in buying these tokens from NVIDIA. So basically all the money NVIDIA invested in Coreweave returned to income.
NVIDIA has concluded similar cloud computing transactions with other so-called Neo-Cloud companies. According to information about information, NVIDIA has agreed to spend $ 1.3 billion this summer in four years, renting about 10,000 of its AI tokens from Lambda, which, as Coreweave, manages data centers, as well as a separate $ 200 million deal for renting about $ 8,000 more in an indefinite period.
For those who think there is an AI bubble, the Lambda agreement is clear proof of foam. Those Nvidia chips Lambda rent time back to Nvidia? She bought them with borrowed money heated according to the GPU’s own value.
In addition to the big investment in Openai and Coreweave, the AI minibus manufacturer also has a value of several million dollars in several other publicly sold companies buying their own GPU or working with related chips technology. These include Arm chips, high -performance calculating company Applied Digital, Cloud Services Company Negus Group and Biotech Company Recursion Pharmaceuticals. (NVIDIA has also recently purchased 4% of Intel shares for $ 5 billion. Like ARM, Intel makes chips that are in some cases Nvidia GPU alternatives, but which usually complement them.)
Earlier this month, NVIDIA has also committed to investing £ 2 billion ($ 2.7 billion) in UK AI startups, including at least £ 500 million in the UK data center operator, which will probably use a few money to buy NVIDIA GPU to provide data centers. NVIDIA has also said it will invest in many British starters both directly and through local venture capital firms, and some of which are likely to return to Openi, as directly or through cloud service providers who in turn will have to buy NVIDIA GPU.
2024. NVIDIA has invested about $ 1 billion in AI startups around the world directly or through its own company venture capital, based on Dealroom and Financial Times; This amount increased significantly from what NVIDIA invested in 2022.
How much of this money returns back to the NVIDIA sales form, it is again difficult to determine. Wall Street research firm Newstreet Research estimates that NVIDIA is investing in Openai for every $ 10 billion, with $ 35 billion in GPU purchases or GPU rental contributions, ie approximately 27% of its annual income last fiscal year.
Such a return would certainly make this type of customer funding. However, this is a concern for analysts about the Bubble of AI. Such circular transactions were a sign of previous technology bubbles and often return to persecution of investors.
In this case, the lease agreement, which Nvidia consists of Open, as part of their latest investment, may seem problematic. Renting, using GPU “Openai”, instead of demanding that they actually buy chips, Nvidia saves Open from the fact that they have to account for the accounts for high token depreciation rates, which will ultimately help the Open. But it means that instead, NVIDIA will have to cover these depreciation costs. In addition, NVIDIA will also take the risk of stuck with GPU inventory that no one wants if the demand for workload does not correspond to the NVIDIA CEO of Jensen Huang Pink.
For some market watchers, the latest NVIDIA offers feel similar to excess technology ups. At the Dot Com Bubble, telecommunications equipment manufacturers, such as Nortel, Lucent and Cisco, lent money for beginners and telecommunications companies to purchase their equipment. Just before 2001 When the bubble exploded, the amount of funding for Cisco and Nortel was increased to their customers exceeded 10% annual income, while the amount of the five best -funding of telecommunications equipment manufacturers exceeded 123% of their total income.
After all, the installation of fiber optical cables and switching equipment was significantly exceeded in demand, and when the bubble exploded and many of those customers beaten, telecommunications equipment manufacturers have left a bad debt in their balances. This contributed to the greater loss when the bubble exploded than it would have been different when network equipment companies lost more than 90% of its value over the next decade.
Even worse were companies such as fiber optical giants in the global crossing, which is engaged in direct “income revenue”. These companies have reduced transactions – often at the end of the quarter to reach Topline forecasts – for which they paid money to another company for services, and then the company agreed to purchase exactly the same value equipment. When the bubble exploded, Global Crossing went bankrupt, and its executives eventually paid large legal settlements related to revenue edging.
These are memories of such transactions that have led to, at least increased eyebrows in some NVIDIA surrounding investments. Goldberg, Seeport’s global analyst, said the transactions had a round funding and was an “bubble -like behavior” emblem.
“The action will clearly arouse” circular “concerns,” NVIDIA reporting its most popular investment in Openai Stacy Rasgon, a Bernstein Research analyst, wrote. Of course, this is a long way from concern to the crisis, but as the company’s assessments increase, this distance begins to stop.
This story was initially displayed by fortuna.com
