Depositphotos
Not only skeptics are already talking about the artificial intelligence bubble, but also heads of leading companies in the industry and investors. Despite the huge investments, there are still no sure prospects for profitability.
According to Bloomberg, while JPMorgan Chase and Mitsubishi UFJ Financial Group are leading a $22 billion-plus loan sale to support Vantage Data Centers’ plan to build a huge data center and Meta is getting $29 billion from Pacific Investment Management and Blue Owl Capital to build another, OpenAI CEO Sam Altman sees parallels to the 1990s dot-com crisis as reports The Verge. The company estimates that it will need trillions of dollars to build AI infrastructure in the future.
This unprecedented investment and lending comes at a time when AI has not actually brought any globally significant financial benefits. The Massachusetts Institute of Technology publishes a report that 95% of generative AI projects in the corporate sector did not bring profit. In a study by Stanford University we are talking about 13% of jobs for beginners taken over by AI in vulnerable sectors, again, mostly without noticeable benefit and with young professionals being prevented from gaining the necessary professional experience. Investors and experts are beginning to show concern.
“It’s quite natural for investors to think back to the early 2000s, when telecom companies were perhaps over-expanding and over-leveraged, and we saw significant write-downs on those assets. So the AI boom certainly raises questions about sustainability in the medium term,” says Daniel Soreed, head of US investment banking at Citigroup.
The initial creation of the infrastructure needed to train and run AI models was largely funded by AI companies themselves, including giants like Google, Meta, and others. Recently, however, money has increasingly come from bond investors and private lenders. Many large tech companies used to finance new infrastructure with corporate debt, which was presumably safe due to existing cash flows. Now, much of the debt financing comes from private credit markets.
“Private lending to AI has been at least $50 billion per quarter for the past three quarters. Even excluding the mega-deals from Meta and Vantage, they are already providing two to three times more than the public markets,” said Matthew Mish, Head of Credit Strategy at UBS.
Many of the new data centers are financed by commercial mortgage-backed securities tied not to a corporate entity but to the payments generated by the complexes. According to JPMorgan Chase, this month, the volume of CMBS for AI structures has already increased by 30% to $15.6 billion compared to the total for the whole of 2024.
On August 8, Daniel Soreed and his colleagues at Citigroup published a report on the specific risks to utilities that have increased borrowing to build the electrical infrastructure needed to power energy-intensive data centers. They and other analysts share a common concern about the huge costs now until AI projects demonstrate their ability to generate revenue in the long term.
“Data center deals are 20- to 30-year financing for a technology that we don’t even know what it will look like in five years. We’re conservatively estimating cash flows going forward because we don’t know what they’re going to look like, there’s no historical basis for that,” says Ruth Yang, global head of private market analytics at S&P Global Ratings.
UBS Group notes that the stress has begun to manifest itself in the growth of in-kind loans to technology-oriented private lenders. According to UBS, in the second quarter, income from this in credit companies reached the highest level since 2020 and reached 6%.
However, cash investments are unlikely to stop anytime soon: credit companies are constantly raising capital and it has to go somewhere, and now they see hyper-scalable companies that deal with AI as a long-term infrastructure asset.
Контент сайту призначений для осіб віком від 21 року. Переглядаючи матеріали, ви підтверджуєте свою відповідність віковим обмеженням.
Cуб'єкт у сфері онлайн-медіа; ідентифікатор медіа - R40-06029.