In 2000, when the dot-com bubble was at its peak, AOL was one of the most popular firms in the world. Its advertising income was growing annually, and the internet pioneer had introduced the web to millions of American homes.
AOL’s announcement of the largest deal in American history, a $360 billion (£266 billion) merger with media company Time Warner, was a positive indication of the future of the internet.
Years after the bubble burst, the truth behind AOL’s explosive growth would come to light. American authorities accused the corporation in 2005 of utilizing fraudulent “round-trip” transactions to boost its earnings by covertly paying its consumers to purchase AOL advertising.
Prosecutors said that the business “effectively funded its own online advertising revenue.”
AOL settled the accusations by paying a $300 million penalty.
The dotcom bubble was characterized by these circular deals. In order to finance new networks and increase sales, telecoms and software businesses paid one another. This increased income and created the appearance of growth until the bubble burst.
Critics of the artificial intelligence (AI) field say they are seeing identical trends 25 years later and speculate that a new bubble may be forming.
Spending spree
Nvidia, a semiconductor powerhouse that has become the most valuable business in the world due to the AI boom, said on Monday that it will invest $100 billion in OpenAI, the AI startup that created ChatGPT.
A large portion of the money may eventually go back to Nvidia because the agreement also calls for OpenAI to invest billions of dollars in data centers that are probably going to house Nvidia’s processors.
The investment will be made in phases, with Nvidia contributing more money as OpenAI increases its expenditures.
The agreement was an investment opportunity, according to an Nvidia spokesperson, and the company was not providing OpenAI with funding to purchase its products. Nothing in the arrangement suggests that it is bad. Analysts acknowledge, though, that it will draw attention.
The deal, according to Stacy Rasgon of the equities research company Bernstein, “will clearly fuel ‘circular’ concerns … and – perhaps justifiably – raise concerns over the legitimacy of the action.”
“Until Nvidia clarifies the appropriate accounting treatment, the optics of such a large investment in a customer will raise questions,” says Vivek Arya of Bank of America.
The largest deal in the AI boom is Nvidia’s $100 billion deal, but it is by no means the only one that has raised concerns.
Earlier this month, Oracle billionaire Larry Ellison momentarily became the world’s richest man with the announcement of a $300 billion agreement with OpenAI, which would include extensive purchases of Nvidia processors.
The agreement basically requires OpenAI to pay Oracle to spend money with Nvidia. Memes circulated online on Tuesday stated that the three corporations had created a “infinite money glitch” – a scenario in which a fault is exploited to generate endless revenue.
The AI explosion that has boosted stock market prices is rife with mutual transactions. Amazon has spent billions of dollars in OpenAI competitor Anthropic, which trains its systems mostly on Amazon’s data centers.
In addition to supporting a number of AI start-ups, Nvidia said last week that it will invest $500 million in British AI firms Wayve and Nscale, both of which use its processors.
Earlier this month, the Dutch tech giant ASML, whose equipment are essential to the production of microchips, announced that it will invest €1.3 billion (£1.1 billion) in the French artificial intelligence start-up Mistral.
AI businesses are investing heavily in data centers and the semiconductors that enable them, resulting in an incredible investment boom. Data center investment is expected to increase by 42 percent this year to $474.8 billion, according to Gartner.
The extent of the spending spree can influence the trajectory of economies. Deutsche Bank analysts predict that without the data center growth, the US “would be close to, or in, recession this year.”
“AI machines—in a very literal sense—seem to be saving the US economy at the moment,” they said.
Considering the high expectations for the technology, the huge expenditure might not be shocking. The CEO of OpenAI, Sam Altman, stated on Tuesday that AI may be able to treat cancer if it acquires the 10 gigawatts of processing capacity that were promised as part of the Nvidia investment.
‘Hope and pray’
However, it is unlikely that there would be any financial return on the investment.
OpenAI generates $12 billion in revenue annually. According to S&P, global generative AI revenues for all companies will reach $30 billion this year. Even though it will reach $85 billion by 2029, this is still far less than the investment cost.
Aswath Damodaran, a professor of finance at New York University, thinks that too much money is being spent on AI infrastructure, considering that the market for AI goods and services is still in its infancy.
The management consultant Bain claims that even in the best-case scenario for AI adoption, when the technology takes over significant amounts of companies’ R&D, marketing, and sales budgets by 2030, revenues would still fall $800 billion below of what is required to finance anticipated infrastructure expenditure.
And the bright scenario may not come true. Last month, researchers at the Massachusetts Institute of Technology stated that 95 percent of organizations that have used AI were seeing “zero return.” The report triggered a brief market panic, wiping $1 trillion off US tech stocks.
Researchers from the consulting firm BetterUp Labs and Stanford University discovered this week that AI could really be decreasing productivity. According to the survey, office workers were employing AI to produce a lot of “workslop,” or polished reports and presentations that are devoid of any true substance. After that, colleagues will have to spend hours fixing that trash.
The tech portfolio manager at T Rowe Price, Dom Rizzo, is not in agreement.
He believes that AI has the potential to be the most significant productivity boost to the global economy since electricity. He does, however, point out that historically, productivity cycles have been accompanied by speculative bubbles.
According to Rizzo, the recent surge in circular agreements and the massive increase in debt used to fund data centers indicate that we are about to reach “the next phase of the bubble.”
These deals have been typical of past market manias, such as the Celtic Tiger real estate bubble in Ireland and the surge in energy trading at the beginning of the new millennium, which crashed along with Enron.
Altman has warned about an AI bubble himself. Is the current state of AI overexciting investors as a whole?, he stated, “Yes, in my opinion.”
Not everyone is quite that concerned. Dan Ives, a Wall Street analyst who is well-known for his consistently optimistic predictions, referred to the Nvidia acquisition as watershed moment.
He stated that although concerns have been raised regarding a “AI bubble” and inflated valuations, we still see this as a 1996 moment for the IT industry rather than a 1999 one.
In such scenario, mark 2028 on your calendar.
