The AI boom keeps accelerating. So why are experts more nervous than ever?

The AI boom keeps accelerating.

The global tech industry is in one of its most explosive phases in decades. Nvidia has reported quarterly profits close to 32 billion dollars, a figure that would have sounded impossible even two years ago. Its market value recently crossed 5 trillion dollars before easing back. Microsoft, Apple, Google and Amazon all sit comfortably in trillion-dollar territory, together generating more than 110 billion dollars in profits in just one quarter. It is, by almost any measure, a period of extremes.

Yet behind the celebrations, a very different mood is building. Economists, investors and even senior executives inside the industry say this wave of spending is starting to feel precarious. The worry is not about whether AI matters, but whether the financial structure around it can support the scale of bets now being made, the New York Times reported.

Much of the anxiety is driven by the numbers emerging from the AI start-up world. OpenAI is now estimated to be worth about 500 billion dollars even though it does not expect to be profitable until around 2030. Anthropic carries a valuation near 183 billion dollars and continues to burn cash. New entrants such as Thinking Machines Labs have reached valuations in the tens of billions within months of launching a first product.

The spending is just as dramatic as the valuations. Anthropic has announced plans to invest 50 billion dollars in new data centres. OpenAI has talked about long-term commitments of up to 1.4 trillion dollars in computing power, anchored in massive projects like the Stargate complex in Texas. In today’s money, Stargate alone would cost several times more than many of the largest infrastructure projects in history.

A large part of this build-out is being financed with borrowed money. Analysts expect companies and governments to pour nearly 3 trillion dollars into data centres by 2028, with close to 1 trillion dollars funded through debt raised from banks and private lenders. Because this borrowing is spread across many institutions, it is hard for regulators and investors to see where the real risks sit.

At the same time, some of the headline deals in AI are financially circular. OpenAI receives billions from partners such as Microsoft but also spends billions buying cloud computing from those same partners. Nvidia has announced huge investments in companies that are, in turn, obligated to spend heavily on Nvidia’s chips. Goldman Sachs has estimated that a significant share of Nvidia’s expected sales next year will come from such intertwined arrangements. Critics argue that these patterns can make the market look stronger and more self-sustaining than it really is.

For now, consumer excitement around AI is easy to see. Hundreds of millions of people use chatbots and image generators and many are willing to pay monthly subscriptions. The real test, however, lies with businesses. Corporate customers are the ones who ultimately decide whether AI becomes a durable revenue engine or remains an expensive experiment.

Surveys show that most companies have tried AI tools, but a similar proportion say they have not yet seen a meaningful improvement in profits. That gap makes investors nervous. If returns do not ramp up fast enough, the enormous cost of training and running AI systems could overwhelm even well-funded players, and heavily indebted firms further down the chain would be especially vulnerable.

Even leaders inside the industry recognise the risks. Alphabet chief executive Sundar Pichai has said publicly that some valuations and spending patterns are driven by irrational behaviour, and that no major company would be immune if there were a sharp correction. The comparison many people reach for is the dot-com boom of the late 1990s, when hundreds of start-ups disappeared almost overnight once the bubble burst, dragging down suppliers and investors with them. Yet the same period also produced lasting giants such as Amazon and Google.

AI may follow a similar path. As Sam Altman of OpenAI has put it, investors may indeed be overexcited and yet still be right that AI is one of the most important technological changes in a very long time. For now, the world’s biggest tech firms are racing ahead, building vast infrastructure and placing century-scale bets. Whether this becomes the foundation of a new economic era or a painful case study in excess will depend on how quickly AI proves its value outside the hype.