Information and data are now emerging about the growing flow of debt used to finance the artificial intelligence (AI) boom and how major corporations are devising financial mechanisms aimed at trying to escape the consequences if the expanding financial bubble bursts.
Initially, after OpenAI launched ChatGPT three years ago, the money provided to finance the construction of the AI data centers came from the cash reserves of the major tech companies.
The Financial Times (FT) has reported that this year US companies have sold $1.7 trillion of investment-grade bonds this year, close to the record of $1.8 trillion in 2020 when there was a rush to take advantage of the ultra-low interest rates at the start of the pandemic.
According to the report, there has been an AI borrowing boom accounting for 30 percent of investment grade issuance “as Big Tech groups including Meta [the owner of Facebook] and Alphabet [the owner of Google], Amazon and Oracle tapped bond markets to fund data centers and the energy systems needed to power them.”
Debt issuance is expected to increase markedly next year with what has taken place so far described as the “tip of the iceberg.”
JP Morgan has said that $1.5 trillion in AI-related debt will be needed by 2028, with AI infrastructure spending projected to reach $5-7 trillion by the end of the decade.
Major tech companies, including Meta, Elon Musk’s xAI, Oracle and data center operator CoreWeave are leading the way in devising means by which they are shielded from a collapse of the boom by setting up special purpose vehicles (SPVs) funded by Wall Street investment firms.
Financial firms, including Pimco, BlackRock, Apollo, Blue Owl and banks such as JP Morgan have supplied at least $120 billion in debt, according to the FT.
The advantage for the firms creating the SPVs is that the debt they incur is off balance sheet making it easier for them to continue raising money in the corporate bond market. But it creates new levels of risk under conditions where it is far from sure who will be on the hook if the projects funded by the SPVs do not generate sufficient revenue.
As the FT noted the SPV structures “increase the danger that financial stress for AI operators in the future could cascade across Wall Street in unpredictable ways.”
It cited one banker “close to data center financing deals” who said two risks were becoming intertwined: the growing risks in private equity markets more generally and the specific risks in the AI boom.
The practices which played a major role in the 2008 financial crisis are also making a return with securitization of AI debt. This is the practice where lenders pool loans and sell off slices of them to other investors in the form of asset-backed securities. So far, the numbers are not large, but they are a warning sign.
The development of AI, if it were used in a consciously controlled and rational manner as augmented intelligence, could provide enormous economic advances. But its development under the capitalist system of private ownership and market relations contains within it the seeds of a financial crisis which could rapidly germinate.
These arise from insufficient revenue being generated from the application of AI to finance the massive spending on data centers, problems associated with the development of the electricity needed to operate them and the prospect that technology changes will lead to more efficient and less costly methods being developed, meaning that the present projects will become so-called “stranded assets.”
Already there are concerns that the circular character of the current deals and agreements has created an artificial boom. In a post on the Institute of New Economic Thinking site, Dutch economist Servaas Storm outlined its extent.
“Crucially” he wrote, “most of the mega-financing deals are remarkably circular. To give you a flavor: Nvidia invests in OpenAI and OpenAI is looking to buy millions of Nvidia’s specialized chips. OpenAI buys computing power from Oracle which buys Nvidia’s GPUs [graphic processing units]. Nvidia owns 5 percent of CoreWeave and sells chips to CoreWeave. CoreWeave’s biggest customer is Microsoft, which is an investor in OpenAI, shares revenue with OpenAI, buys chips from Nvidia and has partnerships with AMD. AMD, a rival to Nvidia, was so eager to land OpenAI as a customer that it issued warrants for OpenAI to buy 10 percent of AMD at a penny a share. OpenAI is a CoreWeave customer and also a shareholder. Nvidia has invested in xAI and will supply it with processors. And so on and so forth.”
The interconnected character of the finance deals means that problems in one area of the AI boom can rapidly spread to others.
Storm said that to many observers “these astronomical circular financing deals” brought back “traumatic memories of the circular financing arrangements of the late 1990s, when vendors reinforced each other’s dotcom stock valuations without generating any real value.”
In fact, the situation is potentially far more serious because the financial system, along with debt, has grown by leaps and bounds since then and has become ever more integrated such that a crisis or collapse in one area of the market has the potential to set off a broader crisis.
The crisis of 2008 and the market freeze of March 2020 were both warnings of such an event.
While there are clear indications that the financing of AI is setting up the conditions for a crisis, there is another issue which must be considered.
What if the estimates of the financial benefits of AI, on which the massive investments have been based, prove to be correct, or at least partially so?
The only way that can happen is if the companies using AI generate vastly increased profits by a massive reduction in their cost structure through the elimination of vast swathes of jobs, particularly in so-called white collar and computational occupations. There are already indications of that starting to take place as major firms announce mass layoffs.
The development of either scenario, or a combination of both, raises the need for the working class to wrest the ownership and control of this major technological advance from the financial oligarchy and utilize it in the development of a planned socialist economy.