Warnings of the AI bubble bursting are increasing

That AI could be a bubble has been circulating for some time. Even Sam Altman and Mark Zuckerberg have already spoken about it. Now Pat Gelsinger has also joined the chorus. The former Intel CEO is sure it's a bubble but believes it will take time to burst—years even. Banks, business media, analysts, and investors are all currently trying to gaze into their crystal balls.

The Bank of England warns, according to The Guardian, is one of many financial institutions warning that a sudden market correction could occur soon. While not explicitly mentioning a burst, it suggests that current valuations and prices cannot be sustained long-term. A correction, after all, always implies that something is wrong. This followed a similar warning from Deutsche Bank. Deutsche Bank is primarily concerned about the US economy, stating that the AI boom is unsustainable. However, current investments in AI are so massive that they are preventing the US from falling into a recession.

The Harvard Economist writes that 92 percent of US GDP growth is based on the boom surrounding AI data centers. At the same time, this is masking economic stagnation. The consulting firm Bain & Company believes that AI companies cannot earn enough to cover their computing costs. An annual revenue of two trillion US dollars would be required to finance the computing power needed to meet the AI demand expected by 2030, according to a report, and it adds: “Even with savings from AI, the world still needs an additional $800 billion to keep pace with demand.”

Meta and Google as winners of the burst

AI companies have all announced massive investments in data centers. Meta plans to build a 2-gigawatt data center as large as half of Manhattan. Project Stargate, announced by Donald Trump as his success, is expected to cost 500 billion, plus a series of further deals, primarily with OpenAI, where affiliations are not always clear.

Furthermore, AI companies are valued extremely highly. OpenAI is now valued at 500 billion. In reality, however, the company is making losses and does not expect to be profitable until 2029 at the earliest. How this will be achieved is completely unclear, as there is no known, foreseeable business model. Such imbalances were partly responsible for the bursting of the dot-com bubble in 2000.

However, the situation is different for Meta and Google, both of which can finance AI developments from their other businesses, primarily advertising. Zuckerberg therefore also assumes a bubble, the bursting of which, however, will make him a winner. Big Tech can then cheaply acquire startups, ideas, and experts.

Nvidia believes in AI

Among the few people who do not expect a bubble is, unsurprisingly, one of the main beneficiaries of the AI hype: Jensen Huang, CEO of chip developer Nvidia, sees no similarities to the dot-com bubble that many fear. On the contrary, he recently said he regrets not having invested more in Elon Musk's AI startup, xAI. This is also his evasive answer to the question in an interview with CNBC how he views the current circular financing models. This concerns Nvidia's investments in AI companies, which are obligated to buy chips from Nvidia with this money.

The fragility of the market was also demonstrated by the emergence of Deepseek at the beginning of the year. In a single day, Nvidia's stock plummeted by 17 percent, and its market capitalization fell by almost 600 billion US dollars to 2.9 trillion. The Chinese AI model was reportedly trained much more cheaply and delivered nearly identical results to ChatGPT. It wasn't quite that simple, but even large tech companies are precisely interested in providing the best models as cost-effectively as possible.

(emw)