As the surge in the US stock market continues, concerns are starting to develop that the bubble based on the massive investment in AI could burst, as parallels are drawn with the dot-com collapse at the start of the century.
A market collapse today would not be simply a repeat of that experience, significant as it was, because the amounts of money involved are so much greater.
Fears of a possible meltdown have been heightened by the recent deal between the leading AI chipmaker Nvidia and OpenAI, which sparked the AI surge with the release of ChatGPT towards the end of 2022.
Under the deal, Nvidia is to invest up to $100 billion in OpenAI to assist it in building massive data centres using Nvidia chips.
The deal is raising questions about the way in which Nvidia is making investments to finance the AI boom to ensure that companies are locked into the use of its chips.
As a recent article on Fortune put it, the “massive data centre build out” has “added to a growing sense of unease that there is a dangerous financial bubble around AI, and that the revenues and earnings math underpinning the valuations of both public and private companies in the sector just doesn’t add up.”
The Financial Times reported that just hours after the Nvidia-OpenAI deal was announced, the global consultancy firm, Bain, released a report which said that AI companies would need to spend $500 billion annually on capital investment to meet anticipated demand. Funding that expenditure would require $2 trillion in annual revenues, but the industry would miss that target by $800 billion.
The scale of the investments is unprecedented. The Nvidia-OpenAI deal calls for “at least” 10GW of computing power. According to the International Energy Agency, 10GW in AI data centres would consume as much energy as consumed in a year by 10 million typical US households.
But there is no certainty about how this massive outlay will be recouped. Last year, OpenAI recorded a loss of $5 billion on $3.7 billion in revenue. This year, according to a report by the business channel CNBC in August, revenue is on track to pass $20 billion. But this is not enough to put the company in the black, and losses are expected to continue.
At that time, CEO Sam Altman told CNBC on the release of ChatGPT-5, “As long as we’re on this very distinct curve of the model getting better and better, I think the rational thing to do is just be willing to run the loss for quite a while.”
In other words, the investment is a massive gamble. Moreover, under conditions where a host of companies are investing in AI, above all the major tech giants Meta, Google, Microsoft and Amazon as well as Nvidia, not all of them will be able to make the sort of profits needed to pay for their investment outlays.
The Nvidia strategy, as exemplified in its deal with OpenAI, is to ensure that however the market develops, its chips, graphic processing units (GPUs), will be at the centre of AI development. Before the latest deal was announced, it had already entered into similar smaller agreements with smaller companies.
The deal with OpenAI is a qualitative leap. Concerns have been raised that it involves “circular” financing of the kind developed in the dot-com bubble, which led to billions of dollars in losses when it collapsed.
Under the agreement, Nvidia will lease its GPUs to OpenAI rather than selling them outright. This means that OpenAI will not have to bear the charge for the high depreciation rates on the chips as new, more powerful ones are developed.
As the Fortune article noted, this means Nvidia will ultimately have to bear the depreciation costs. “What’s more,” it continued, “Nvidia will also take on the risks of being stuck with an inventory of GPUs no one wants if demand for AI workloads don’t match Nvidia CEO Jensen Huang’s rosy predictions.”
The Nvidia arrangements bear a close resemblance to those engaged in by telecom equipment makers 25 years ago. Firms such as Nortel, Lucent and Cisco lent money to telecom companies. But the bubble collapsed because the supply of equipment exceeded the demand, and the networking companies lost as much as 90 percent of their value over the next decade.
In a comment to Fortune, Jay Goldberg, an analyst at Seaport Global, said the Nvidia deals had a whiff of circular financing and were symptomatic of “bubble-like behaviour.”
Stacy Rasgon, an analyst at Bernstein Research, wrote in a note to investors that the Nvidia-OpenAI deal would “clearly fuel ‘circular’ concerns.” He said that while it was a long way for concern about a crisis, the distance was starting to close.
There is also a macroeconomic dimension to circularity. According to calculations by Harvard economist Jason Furman, reported by the FT, investment in processing equipment and software comprises some 4 percent of GDP and was responsible for 92 percent of growth in the first half of the year.
Investment in AI and other technologies, accompanied by a share market surge, is being carried out because it is claimed the economy is “strong.” But outside that investment, growth is essentially zero.
This situation was underscored by a report from the payroll assessor ADP yesterday, which said private sector employment in the US fell by 32,000 last month, the largest drop in two and a half years, after economists’ predictions were for a 50,000 increase.
The Nvidia-OpenAI deal needs to be viewed within the framework of the speculative binge, powered by ultra-cheap money which has powered the rise of Wall Street since the 2008 global financial crisis. The S&P 500 index is at around 6,688. At its nadir after the crisis, it was 666 in March 2009.
There has been a 100-fold increase in the index since then, underlining the growing divorce between the stock market and an underlying real economy on which it ultimately depends. The growth of US GDP over the same period has been from $14.48 trillion in 2009 to $30.5 trillion today—little more than double.
Moreover, the rise in the market has been increasingly concentrated in the tech giants—the so-called Magnificent Seven. Together, Alphabet (Google), Amazon, Apple, Meta (Facebook), Microsoft, Nvidia and Tesla account for around 37 percent of the S&P 500 market capitalisation, the highest share on record with Nvidia playing the leading role.
Since October 2022 and the start of the AI stock market surge, its shares have risen 1,350 percent. The development of AI has the potential for a massive boost in the productivity of human labour and the advancement of civilisation. But its exploitation within the framework of capitalist social and economic relations, based on private profit, is sowing the seeds for a major financial crisis, which, as the experience of 2008 revealed, will have devastating economic and social consequences.