Top economist warns of AI bubble: companies earn nothing from it

American companies are investing billions of dollars in artificial intelligence (AI), but the financial benefits remain largely unavailable for the time being. This is according to Torsten Slok, chief economist at investment company Apollo Global Management. His analysis shows that large technology companies in particular benefit, while the rest of American business is making little progress. The warning comes at a time when concerns about a possible AI bubble on the financial markets.

Slok examined the financial performance of the seven largest American technology companies, also known as the Magnificent Seven. He compared these companies with the remaining 493 companies in the S&P 500, a major US equity index.

The difference is striking. At the major technology companies, revenue per employee increased from about $550,000 in 2022 to almost $800,000. For the other companies, this amount remained virtually unchanged at around $550,000.

The profit margins also show a similar picture. While the largest technology companies are improving their profitability, the margins of the other companies remain stuck at around ten percent.

Slug sees this as an important warning signal. Despite all the investments, he believes there is still no clear evidence that AI is accelerating productivity growth in the broader economy.

The economist emphasizes that the ultimate success of AI is not determined by a handful of technology giants. It is precisely the hundreds of other listed companies that have to demonstrate that the technology actually generates more turnover and profits.

However, there is an important nuance to this. The figures show that broad financial benefits are lacking for the time being, but do not prove that individual companies do not earn anything from AI.

Another problem is the increasing cost of using advanced AI models. Companies often pay per amount of processed text or information, expressed in so-called tokens.

Data from payment platform Ramp shows that the ten percent largest users are responsible for 95.6 percent of spending on OpenAI and Anthropic. Especially companies that are financed with venture capital spend large amounts of money.

Traditional companies are considerably more reluctant. According to Slok, more and more companies no longer regard AI spending as an investment that generates growth, but as a cost item that puts profitability under pressure.

Some large employers are now intervening. Consultancy firm Accenture limited employee spending on AI model Claude to $29 per month this month.

Goldman Sachs also seems cautious. CEO David Solomon stated last week that his bank has no need at all for the most advanced AI models for the vast majority of its operations.

That's striking, as Goldman Sachs has simultaneously announced an AI investment of about $500 million.

The findings appear as investors increasingly question whether the huge investments in AI can eventually be recouped.

Elon Musk warned last weekend about the risks of investing in companies that offer computing power for AI. According to him, fierce competition and financial problems can lead to bankruptcies and a wave of acquisitions.

CoreWeave in particular is under a magnifying glass. The company has accumulated more than $50 billion in debt and is consuming large amounts of money to expand its operations, according to reports.

Not everyone shares the gloomy expectations. Jon Gray, president of investment firm Blackstone, dismissed warnings of an AI bubble last week. He described the demand for AI as growing almost exponentially.

This week, investors will get new information when Alphabet, Amazon and Microsoft publish their quarterly figures. In particular, it will be examined whether their huge AI investments actually generate additional income.

Research firm Gartner warned earlier this month that $3.5 trillion in investment in AI infrastructure may need to be written off by 2030.

The costs of AI developer Anthropic are also in the spotlight. The company reportedly spent $80.7 billion in the first half of this year and has taken steps toward an IPO, according to reports.

For Slok, development outside the technology sector remains decisive. As long as AI spending is concentrated in a small group of companies, he expects the impact on the broader US economy to be limited.