Why the AI Bubble is Worse Than You Think

In a few years, artificial intelligence went from being a technological promise to a global economic race. Ever since the AI boom took off, there have been warnings about a bubble. Tech companies are pouring billions into advanced chips and data centers. And the final bill may run into the trillions. But behind the technological enthusiasm, signals that worry some economists have begun to appear. That's led to wider market nerves again about the massive spending on AI infrastructure and how much of it is being funded by debt. For example, OpenAI has signed infrastructure commitments that exceed trillions of dollars for the coming years. OpenAI, the maker of ChatGPT, has signed a seven-year, $38 billion deal to buy cloud services from Amazon. Meanwhile, the tech industry is in a race to build increasingly larger data centers. Policymakers are paying a lot of attention to this issue and the demand created by the construction of new data centers. But to sustain these projects, companies in the sector would have to multiply their revenues several times before the end of the decade. OpenAI essentially needs to grow revenue by 85x in five years to have even the remotest chance of making the bills. Announcements continue and investments keep growing. But more and more analysts are starting to question whether this expansion can really be sustained. So, why the AI bubble is worse than people think? Over the past few years, artificial intelligence has become one of the main investment drivers within the technology sector. Companies around the world began investing enormous amounts of capital in data centers, model development, and specialized hardware. And the numbers show the scale of this phenomenon. Some estimates indicate that global spending on artificial intelligence could exceed $500 billion per year before the end of the decade. At the same time, market enthusiasm has driven extraordinary valuations for companies dedicated to developing this technology. The company's record sales have been fueled by an emerging technology. We're the world's engine for AI. Here's how an under-the-radar graphics processor is cashing in on the AI craze. This has led several companies in the sector to reach valuations above $100 billion in just a few years. However, the rapid growth has also raised doubts among some financial analysts. Economic history shows that when expectations exceed real revenues, bubbles can form, such as when the crypto market reached $3 trillion. For this reason, more and more experts are questioning whether the growth of artificial intelligence is sustainable. These projects are set in stone. This money will be spent. The next 12 months will continue to see this very strong capex cycle. But sustainability of that is a cause for concern. The development of advanced AI systems requires gigantic infrastructure. The most complex models require data centers with thousands of processors working at the same time. Some already exceed 10,000 chips. For this reason, several technology companies have begun signing financial commitments that exceed any recent precedent within the industry. One of the most striking examples is the case of OpenAI, which has signed infrastructure agreements close to $1 trillion for the next five years. In addition, the cost of building specialized facilities continues to increase. Amazon, for example, is one of the main companies increasing spending on data centers across the country. The company is set to put $10 billion more dollars into Ohio over the next five years, expanding its Amazon Web Service data centers, housing equipment for powering AI, machine learning, and other cloud computing technologies. This growth has also boosted the energy market needed to power these facilities. In some individual projects, electricity consumption can equal that of cities with more than 100,000 inhabitants. This has made the scale of these projects reflect the level of However, this level of investment also creates enormous pressure on the revenues companies will have to generate in the future. To make all of this math work, Bain estimates that $2 trillion of new revenue is needed. That is 7% of U.S. GDP. To sustain such large financial commitments, companies need to expand their business at an extraordinary speed. The expected growth exceeds what most technology companies have achieved in the past. For example, open AI would have to go from about $12 billion in 2025 to nearly $1 trillion annually by 2030. A leap of this magnitude would be one of the fastest corporate growths recorded in modern history. In addition, the challenge is not limited to increasing revenue. At the same time, training advanced models can cost hundreds of millions of dollars per project. The thing that's happened in the industry since, is we're now in a world where we made incredible progress with distillation, we learned a lot to do small models, and these reasoning models in particular can be, it's not cheap, it's still expensive to train them. Given this situation, the artificial intelligence services market is still in an early stage. Some estimates indicate that less than 20% of global companies use these tools intensively. This has forced many companies to bet on extremely accelerated growth. However, that expansion also depends on a complex network of financial relationships within the technology sector itself. There are dozens of investment structures just like this. And the problem is these Ouroboros-like dealings are impossible to track and measure. Because the revenue of one company is the cost of another company is the investment of another company. Within the artificial intelligence ecosystem, many companies maintain investment and consumption relationships at the same time. Thus, tech companies create an economic network where money constantly circulates among them. A clear example appears in the relationship between Microsoft and OpenAI. Microsoft has invested more than $13 billion in the company to boost the development of its models. In turn, OpenAI uses Microsoft's cloud computing infrastructure to train its artificial intelligence systems. The company Microsoft first invested in in 2019 on the condition that they were their exclusive partner infrastructure-wise with their Azure cloud. In return, Microsoft got certain access to some of this OpenAI tech. In addition, Microsoft uses part of that revenue to buy chips from NVIDIA, a market that already exceeds $500 billion. These chips have become the central component of modern artificial intelligence systems. That is why demand for these processors has been so strong that NVIDIA's revenues grew more than 200% in just two years. This type of relationship creates an economic circuit where several companies become investors, customers, and suppliers at the same time. A circular deal is when companies are basically blowing money, products, services in between two of them. For this reason, understanding the real value of the sector is difficult. Since more than 40% of revenues come from transactions between the same companies. When several companies buy products and services from each other, it becomes more difficult to identify which revenues come from independent market demand. In some cases, the revenue of one company represents at the same time the operating expense of another company within the same technological ecosystem. Recent studies indicate that more than 60% of cloud spending on artificial intelligence comes from large technological companies within the sector itself. This means that a significant part of demand comes from the same companies trying to expand artificial intelligence infrastructure. So the way we have to think about this is from the entire value chain perspective, that it's not just about mega cap tech names. There's other semiconductor companies, there's other data center companies, there's other software companies that are benefiting from the rise of this theme. Despite the similarities, some economists point out that the current situation is not identical to other technological bubbles of the past. For example, during the internet boom in the year 2000, many companies reached enormous valuations without generating significant revenue. In comparison, today large technological companies produce gigantic profits every year. Companies like Microsoft or Apple generate tens of billions of dollars in annual profits. Yeah, and certainly we're advancing today on the strength of the Mag7. So, so far a year to date, it had been Meta and Microsoft and Amazon back in February, NVIDIA had made new all-time highs. However, the current problem does not lie only in the profitability of these companies. Today, some artificial intelligence companies have seen their value rise by more than 300%. This shows how high expectations are about the future of AI. But it also raises an important question about who would absorb the losses if those expectations are not met. People are worried specifically when it comes to comparisons to the dot-com bubble of assets that have vastly exceeded their value, and then if things go up and there's no revenue to support it, things go down. Even some statements within the sector itself have generated debate. The chief financial officer of OpenAI mentioned that artificial intelligence infrastructure projects might need government backing in the future. For example, in 2008, several banks received public support worth billions of dollars to avoid collapse. The idea of public backing implies that if the business model does not generate sufficient revenue, governments could be pressured to intervene. This would open the door to a scenario where citizens end up assuming part of the financial risk created by the technology sector. And this is already being felt, as support for strategic companies already exceeds $1.2 trillion. President Trump, he recently just announced the Stargate project, which is a collaboration among OpenAI, Oracle, and SoftBank that aims to invest up to $500 billion over just four years to develop AI infrastructure in the United States. While AI is transforming many industries thanks to its ability to automate tasks, the current rate of investment seems unsustainable, as many companies would need to reach unprecedented revenue levels to become profitable.