AI: Money-making machine or a billion-dollar sinkhole?

Artificial intelligence chipmaker Nvidia last week said sales had reached a higher-than-expected $30 billion (€27.03 billion) in the last quarter, though added that growth was slower than the furious pace seen in previous quarters.
Still, shares in the company dipped about 5% in after-hours trading following the report. Even though sales and profit, which hit $16.5 billion in the period, more than doubled from a year earlier, investors showed nervousness that Nvidia’s extraordinary growth, spurred by the AI frenzy, may be showing signs of easing.
“Such a massive amount of money has gone to tech and semiconductors in the last 12 months that the trade is completely skewed,” said Todd Sohn, an ETF strategist at Strategas Securities, in a note to investors.
The sums of money currently being invested in AI companies are enormous. US investment bank Goldman Sachs expects an AI investment volume of around $158 billion this year, with about half of that amount going to the United States. In a June research report titled GEN AI: Too much spend, too little benefit? Goldman said “tech giants and beyond are set to spend over $1 trillion on AI capex in coming years.” 
These funds would flow into significant investments in data centers, chips, other AI infrastructure and the power grid. Whether these massive investments will ultimately generate returns beyond the current “picks and shovels” phase, however, remains unclear.
But for major tech companies, withdrawing from the AI race is not an option. During the presentation of the latest financial results of Google parent company Alphabet, CEO Sundar Pichai said “the risk of underinvesting in AI infrastructure is dramatically greater than the risk of overinvesting.”
Facebook parent company Meta appears to view AI’s potential in the same way, as its spending on the technology also remains high, rising to over $24 billion last quarter. Meta expects AI spending of between $37 and $40 billion this year, and is preparing investors for a “significant” increase in 2025,” German news agency dpa reported.
Leopold Aschenbrenner, a former employee of AI pioneering company OpenAI who was fired for disclosing classified company documents, wrote in a June 2024 research paper that the boom is “investment-led,” but that it is taking time to train AI, build chip factories and develop energy infrastructure. Profits will come later, he wrote, but companies are already generating good revenues now.
Currently, about 27% of companies in Germany use AI, said Klaus Wohlrabe, head of surveys at Munich-based Ifo Institute. Some 17% plan to use AI in the coming months. “The trend is likely to pick up more speed,” he told DW.
Wohlrabe, however, also said the think tank’s surveys “do not show the extent to which business processes are fundamentally changed by generative AI,” and that “this is just beginning.”
Christian Temath from an initiative called KI NRW, which seeks to promote AI use in the German state of North Rhine-Westphalia, said practical applications that lead to greater efficiencies in companies and large-scale productivity gains have yet to emerge.
“I don’t think every billion currently being spent on computing capacity in the US will be recouped one-to-one,” he told DW.
Rita Sallam, an analyst at US market research firm Gartner, believes that following last year’s AI hype, executives are “impatient” to see returns on AI investments. “Yet organizations are struggling to prove and realize value. As the scope of initiatives widen, the financial burden of developing and deploying GenAI models is increasingly felt,” she said.
 Gartner predicts that at least 30% of AI projects will be abandoned after proof of concept by the end of 2025, due to “poor data quality, inadequate risk controls, escalating costs or unclear business value.”
Jim Covello of Goldman Sachs has also warned that despite its high costs, the technology is far from being useful. “Over-building things the world doesn’t have use for, or is not ready for, typically ends badly,” he said in the June report. Venture capital firm Sequoia Capital and hedge fund Elliott Management share a similar view, suggesting that tech companies are already “in bubble territory.”
To describe the development of breakthrough technologies like generative AI, Gartner’s so-called hype cycle is often cited.
First, a potential technological breakthrough is announced and celebrated in the press, although no viable products exist yet. Exaggerated expectations lead to hype. Then comes the trough of disillusionment, as initial products are not as successful as expected. Next, new applications emerge that succeed in the market. The development stabilizes on the plateau of productivity when mainstream applications are running.
Applied to generative AI, the release of ChatGPT in November 2022 triggered the hype. It seems clear we have not yet reached the plateau of productivity.
A collapse of the hype was feared in early August when, among other things, shares in Nvidia plummeted and then again in early September, when the chipmaker shed nearly $280 billion in market value in one day.
The AI race continues, however. How long it will last and whether it will be successful is unknown, as not all hyped technologies make it out of the trough of disillusionment.
Recently, more voices have suggested the AI hype might be a bubble. And bubbles have the unpleasant tendency to sometimes burst, causing significant turmoil in financial markets.
Experts from the rating agency Standard & Poor’s believe the path to monetization and maturity for AI will be “longer than previously expected.”
“By far the biggest beneficiary of AI spending by companies is Microsoft,” the S&P experts said in August.
The number customers of Microsoft 365 Copilot, a generative AI chatbot, has increased by more than 60% compared to the previous quarter, and the number of daily active users has doubled. Goldman Sachs analyst Sung Cho believes there could be “a pause in the near term” which is going to “dictate the shorter-term direction of markets.” What he called killer applications that justify the massive investments have yet to be invented.
Brook Dane, also a Goldman Sachs analyst, said investors will “need to see, at some point over the next year to year-and-a-half, applications that use this technology in a way that’s more profound than coding and customer service chatbots.” If it was just that, investors would be “massively overspending on this.”
But Dane and Cho, both portfolio managers on the Fundamental Equity team in Goldman Sachs Asset Management, are convinced AI will be one of the biggest trends of all time, both in the medium and long term.
Daron Acemoglu, a professor at the Massachusetts Institute of Technology, is more skeptical. He estimates that “truly transformative changes won’t happen quickly and few — if any — will likely occur within the next 10 years.
Quoted in the Goldman Sachs report in June, he said “only a quarter of AI-exposed tasks will be cost-effective to automate within the next 10 years, implying that AI will impact less than 5% of all tasks.” 
He also predicted that AI’s productivity effects within the next decade should be “no more than 0.66%,” and an even lower 0.53% when adjusting for “the complexity of hard-to-learn tasks.” That figure, he concluded, roughly translates into merely 0.9% higher gross domestic product for the US over the decade.
This article was originally written in German.