ENTERTAINMENT NEWS
Entertainment news

The 'circular financing' of Artificial Intelligence, in its most difficult hour

Updated

Four years after the launch of ChatGPT, the tech sector's frantic race to lead AI has reached its peak. Doubts about the financial health of the giants are shaking the markets

The OpenAI logo is seen on a mobile phone in front of a computer screen.
The OpenAI logo is seen on a mobile phone in front of a computer screen.AP

What is Artificial Intelligence (AI)? A technological evolution? A 'peak' of the credit cycle - which is just a polite way of saying a bubble - disguised as a technological revolution? Or both? Less than two weeks remain until November 30 marks four years since the AI boom triggered by the launch of the chatbot ChatGPT from OpenAI, which reached one million users in five days and one hundred million in less than two months, according to data from the Swiss banking giant UBS, thus unleashing a historically significant investment wave that may now be coming to an end. The big question is whether this is a correction or the bursting of a bubble that has supported much of the global economy over the past four years. Whatever the answer, it is clear that the market tone has changed. Anthropic, the global leader in AI, went public on Friday with a valuation between $1.3 and $1.4 trillion, $700 billion less than what the market expected just six weeks ago, according to Reuters. OpenAI has canceled its IPO plans. The global leader in AI chip design and production, Nvidia, has lost $1.6 trillion (over ¤1.4 trillion) in the stock market in a month and a half. Alphabet(Google) has lost $1.4 trillion since May.

These figures are starting to approach those of the dot-com bubble, which in 2000 generated technology investments in the U.S. equivalent to 4.5% of GDP, according to data from the International Monetary Fund (IMF). Because the data from the BIS and Goldman only refer to five major companies: Alphabet (Google), Microsoft, Meta, Amazon, and Oracle. They do not include the neoclouds, that is, the new companies specializing in providing AI services in the cloud, such as CoreWeave, nor the major chip manufacturers like Nvidia, AMD, or TSMC, nor the data center manufacturers like Vertiv, nor memory chip manufacturers, including Micron, which rose over 700% in a year, breaking the $1 trillion market cap barrier in July, and whose stock has since fallen 23%. The major hyperscalers have a key point when defending their investments: their financial solidity. Of course, the resources they are allocating, combined with the - at least for now - moderate capacity of AI to generate income comparable to the investment, are impacting their balance sheets. Amazon's free cash flow has gone from being positive at $430 million to negative at $34.837 billion in a year. Alphabet's has gone from $24.461 billion positive to $9.065 billion negative. Free cash flow measures the cash left to a company after deducting investments in fixed assets (i.e., tangible things, not financial assets).

Their shift from black numbers to red indicates that, to finance the AI race, even tech giants need debt and dilution of value. Alphabet's case is a good example. In 2025, it was conducting share buybacks, that is, buying shares with its own money to raise their price. In 2026, it is undertaking what JP Morgan calls the largest capital increase ever carried out by a company in history, expecting to raise $90 billion, almost as much as the entire value of CaixaBank. This amount will be added to a series of bond issuances for another $80 billion. All to finance investments of between $195 billion and $205 billion in AI just this year, which the company plans to "significantly increase" by 2027, which, according to S&P Global Market Intelligence, could mean another $312 billion. The private debt issued to finance AI investment is so substantial that, according to the president of the Chicago Federal Reserve and former economic advisor to Barack Obama, Austan Goolsbee, "it is plausible" that it is complicating financing for states, including the U.S. itself. With a bit of humor - and exaggeration - one could say that AI has already achieved a monumental change: creating more debt than money to invest. Although the circular structure of AI has few precedents, the scale of its investments is not unusual in the early stages of a new technology. The dot-com bubble is a poor historical example because a significant portion of its innovations applied to existing infrastructures or could be modernized without large expenditures. AI, with its data centers, cloud, expensive microchips, and a demand for electricity so enormous that companies like Alphabet, Amazon, or Microsoft are reaching agreements to purchase the output of nuclear plants for decades, is in a different league.

Perhaps the best comparisons are with other technological revolutions, such as the canal mania in the UK in the 1790s, and the railways, also in that country, in the 1840s, as well as with the accelerated electrification process of the industrialized world in the late 19th and early 20th centuries. All three were extraordinarily capital-intensive. Only the canal mania absorbed investment equivalent to 8% of the UK's GDP, then the dominant economy in the world. But canals do not depreciate. Microchips do. And very quickly. That is a huge difference between this cycle and previous ones. According to Goldman Sachs, an electrical grid has a lifespan of a quarter-century or more. An AI GPU becomes obsolete in four years. Maybe sooner. How much will Nvidia chips be worth in the investment vehicle of Apollo that xAI will use in four years? Probably much less than now. If investment in AI slows down, this delicate financial structure may suffer. And, with it, the global economy. Hyperscalers, neoclouds, and other fauna of the new AI ecosystem have driven Wall Street and the dollar upward. Because the volume of investment in this technology is so unimaginably large that it has captured extraordinary amounts of capital from around the world. And, especially, from Europe, which has capital but lacks both financial infrastructure and technological innovation to deploy it. The result is that, according to the U.S. Department of the Treasury, between January and August of this year, Europe - including the UK and Switzerland - invested $255.7 billion (¤226.9 billion) in the U.S. stock market. If there is a bubble and it bursts, we will burst too.