Concerns over the rapid development of artificial intelligence (AI) have led to calls for a slowdown in the industry, prompting debates about potential impacts on the global economy and stock markets. Last week, Dario Amodei, chief executive of Anthropic, urged major AI companies to decelerate their progress, a proposal that gained backing from influential figures including Sam Altman of OpenAI and entrepreneur Elon Musk. However, not all industry leaders agreed; Nvidia’s Jensen Huang and Meta founder Mark Zuckerberg expressed reservations about the suggested pause.

The growing apprehension about AI’s existential risks has united an unusual range of voices, spanning political and ideological lines. Right-wing activist Steve Bannon and socialist Bernie Sanders appeared together to advocate for increased oversight. Both the United States and the United Kingdom are actively considering regulatory measures to curb the use of the most advanced AI systems. Even the British monarch reportedly voiced concerns about the potential dangers of AI at a recent gathering of technology leaders.

Investor confidence has visibly fluctuated amid the ongoing debate. In a single day last week, the combined market value of 25 leading AI-related companies, including Nvidia, Samsung, and Micron, fell by nearly $700 billion before recovering some losses by the week’s end. Analysts suggest that a slowdown in AI development could trigger a significant correction in global stock markets. Joachim Klement, head of research at Panmure Liberum, warned that decreased AI progress could suppress data center expansion, given that training cutting-edge models is among the most capital-intensive activities in the sector. The five largest hyperscalers — major cloud service providers — are expected to invest $5 trillion in AI infrastructure over the next five years, supporting a wide network of suppliers across the semiconductor, electrical equipment, and real estate industries.

Klement also highlighted risks tied to “take-or-pay” contracts between AI developers and hyperscalers. For example, Anthropic has reportedly secured contracts worth over $200 billion, including a $100 billion, ten-year deal with Amazon Web Services. Should AI development slow, these firms might face declining revenues but maintain fixed, substantial cost obligations, a situation that could undermine investor confidence. Both OpenAI and Anthropic have delayed their initial public offerings amid these uncertainties, with Anthropic targeting a valuation near $2 trillion. The firm claims annualized recurring revenues reached $65 billion by July, up from $9 billion earlier this year, reflecting rapid growth of over 700 percent.

Despite this, competition is intensifying. Anthropic’s newer, high-cost model, Fable 5, reportedly accounted for only a small fraction of usage in August, while open-source AI models and OpenAI’s Astra model offer challenging alternatives, sometimes outperforming Anthropic’s offerings on specific tasks. Some analysts view these trends as signs of emerging weaknesses in the AI market, though others remain optimistic about ongoing enterprise adoption and its potential to sustain the sector’s expansion. In the United States and the United Kingdom, only approximately 17 to 18 percent of businesses currently employ AI, indicating significant room for growth.

Experts also emphasize the distinction between frontier AI model development and broader industry adoption. Adrian Cox of Deutsche Bank noted that many businesses benefit from AI applications without relying on the latest, most advanced models. Additionally, industry figures highlight the profitability of AI “inference”—the application layer where AI models are integrated into usable software—rather than just the costly training phase. This inference phase, charged by usage tokens, is expected to become increasingly scalable and lucrative as AI agents capable of autonomous task performance emerge.

In parallel, Nscale, a British data center developer positioning itself as a foundational player in the AI infrastructure ecosystem, recently released a prospectus ahead of a planned initial public offering in New York. The company claims committed revenues exceeding $100 billion and aims for a valuation of around $35 billion. Its board includes notable figures such as Sheryl Sandberg and Sir Nick Clegg. However, much of Nscale’s contract value is not yet realized, with a relatively small active computing capacity and significant revenue concentration in a single customer, Anthropic, which agreed last month to spend about $45 billion on cloud computing power from Nscale. The company has also reported substantial net losses in recent periods and raised concerns over its ongoing viability before management implemented plans to address funding gaps.

The timing of Nscale’s IPO, amid calls from its key client Anthropic for an industry-wide AI slowdown, underscores the complex dynamics facing the AI sector. Investors must weigh the promise of sustained enterprise demand and technological innovation against the risks posed by regulatory scrutiny, competitive pressures, and the capital-intensive nature of AI development.