- Economists are beginning to question whether AI’s promised revolution in productivity and efficiency will materialize. Carl-Benedikt Frey of the University of Oxford noted that “AI adoption has actually declined since the summer,” warning that without sustainable use cases, the hype could give way to a bubble burst.
- OpenAI, the sector’s flagship company, generated $3.7 billion in revenue last year but reported expenses close to $9 billion. Although it expects revenue to rise to $13 billion this year, analysts estimate it could burn through as much as $129 billion by 2029. Mills argues that generative AI firms such as ChatGPT and Elon Musk’s Grok undercharge for their services and must raise prices to become profitable.
- Companies will begin tracking AI’s return on investment (ROI) to ensure it delivers tangible business impact, marking the start of a more disciplined and results-driven era for artificial intelligence.
The global artificial intelligence (AI) boom continues to attract massive investment, with billions being poured into infrastructure, startups, and top talent. Major announcements this year include a $500 billion AI supercomputer partnership between OpenAI, Softbank, and Oracle, and a $100 billion fund by OpenAI and Nvidia to reinforce U.S. dominance in advanced chips.
Meanwhile, Chinese tech giants Alibaba and Tencent have increased funding to accelerate China’s ambition of becoming a global AI leader by 2030. Since ChatGPT’s launch in late 2022, AI-related stocks have added around $17.5 trillion in market value, driving three-quarters of the S&P 500’s gains and pushing firms like Microsoft and Nvidia to record valuations.
However, dw.com reports that, despite the impressive growth, early signs of an AI slowdown are emerging. Corporate adoption of AI tools has started to decline, and spending is tightening as expectations outstrip actual returns.
Economists are beginning to question whether AI’s promised revolution in productivity and efficiency will materialize. Carl-Benedikt Frey of the University of Oxford noted that “AI adoption has actually declined since the summer,” warning that without sustainable use cases, the hype could give way to a bubble burst.
Data from the U.S. Census Bureau underscores this concern, showing that AI-tool usage among large firms fell from nearly 14% in June to under 12% by August. Persistent technical challenges, including hallucinations — when AI systems generate false information — and poor reliability of autonomous agents, have dampened confidence.
Experts argue that current AI systems, unlike human workers, do not learn from experience, and thus require “continual learning” to adapt effectively over time.
Investment enthusiasm also appears to be cooling as financial realities catch up. Venture capital funding for AI startups fell 22% in the third quarter of 2025, though total funding stayed above $45 billion for the fourth consecutive quarter. Economist Stuart Mills from the London School of Economics cautioned that the enormous sums invested in AI far exceed current revenue streams.
OpenAI, the sector’s flagship company, generated $3.7 billion in revenue last year but reported expenses close to $9 billion. Although it expects revenue to rise to $13 billion this year, analysts estimate it could burn through as much as $129 billion by 2029. Mills argues that generative AI firms such as ChatGPT and Elon Musk’s Grok undercharge for their services and must raise prices to become profitable.
Julien Garran of MacroStrategy Partnership offered a stark warning, estimating that misallocated capital in the AI sector equals about 65% of U.S. GDP — four times greater than the pre-2008 housing bubble and 17 times the scale of the dot-com crash.
Such figures suggest the current AI investment surge could be one of the largest speculative bubbles in modern history.
Investor caution is already surfacing in Big Tech earnings reports.
While companies like Palantir, AMD, and Meta reported strong AI-related revenues, their stock prices dropped amid skepticism about long-term sustainability. Mills highlighted the disconnect between hype and productivity, noting that AI’s business applications are still limited to low-value tasks rather than core value creation.
Nvidia remains the key player to watch. With its chips powering much of the AI revolution, Nvidia reported record-breaking sales — $46.7 billion last quarter, mostly from its data center segment — and expects $54 billion in the next quarter. Yet some analysts warn that even Nvidia’s dominance cannot mask the industry’s deeper structural issues.
Experts are divided on what happens next. Gary Marcus of New York University believes most generative AI companies are “wildly overvalued” and predicts an imminent collapse, citing weak technical and economic fundamentals. Garran agrees that AI model development has reached diminishing returns, with escalating training costs and only marginal performance gains.
Others see a softer landing ahead. Sarah Hoffman of AlphaSense expects a “market correction” rather than a full-scale crash. She predicts a maturing phase where businesses focus less on hype and more on measurable outcomes.
Companies will begin tracking AI’s return on investment (ROI) to ensure it delivers tangible business impact, marking the start of a more disciplined and results-driven era for artificial intelligence.
