The rise of AI: Boom, bubble or neither? | Number Theory

.

‘Artificial Intelligence’ became an everyday phrase in 2025, with generative AI tools spreading rapidly across offices, classrooms and consumer products. That pace of adoption has been matched by extraordinary investor enthusiasm, driving funding, valuations and markets across the globe. The question now is whether this reflects the early stages of a durable technological shift, or a bout of exuberance running ahead of fundamentals.

Representational image. (Reuters)

The rise of AI: Boom, bubble or neither?

• 
  A familiar surge

  In the 1990s, the listing of Netscape marked the moment the internet became investable, setting off a sharp rally in the Nasdaq as investors tried to price a poorly understood future. This ended with the “dotcom bubble” bursting in 2000. The public launch of ChatGPT on November 30, 2022 may be a similar starting point. The Nasdaq has risen strongly since, driven by the belief that AI will power the next phase of technology-led growth. To be sure, part of the AI hype is the result of the fact that the firms investing in it are already highly profitable. Unlike the dotcom bubble, the hype is also driven by already established companies rather than startups with nothing to show.
• 
  AI now dominates funding flows

  Global venture funding data shows the share of AI rising sharply over the past decade, from a relatively small slice of investment in the early 2020s to a dominant position by the mid-2020s. By 2025, AI alone accounted for about $202 billion in global venture funding, close to half of all venture capital deployed during the year. The boom has also been highly concentrated geographically. Crunchbase estimates that roughly 79% of global AI funding in 2025 went to US-based companies. Within the US, just the San Francisco Bay Area raised more than three quarters of total US AI funding. PitchBook–NVCA data also shows that AI accounted for 64.3% of total US venture deal value in 2025, up sharply from 9.3% in 2015.
• 
  Valuations racing ahead of revenues

  Companies such as OpenAI, Anthropic, xAI and Mistral AI moved from modest valuations in the early 2020s to tens of billions and, in OpenAI’s case, to $500 billion by late 2025. The spending behind these valuations is enormous. OpenAI has signed a web of multi-year infrastructure deals with major technology partners, including a $38 billion cloud agreement with Amazon for access to hundreds of thousands of GPUs, and a partnership with Nvidia, which will invest up to $100 billion to deploy at least 10 gigawatts of compute capacity. Reports also point to contracts worth hundreds of billions with Oracle, Broadcom, Microsoft, AMD and others. Independent forecasts by venture capitalist Tomasz Tunguz underline the scale of the challenge in revenue terms. To support its spending commitments, OpenAI would need to increase revenue from around $12 billion in 2025 to $983 billion by 2030 even under generous assumptions on margins, bringing it in a league with the largest companies that exist today. Such growth is not impossible, but rare and hard to sustain. It also underlines the exacting expectations in today’s AI valuations.
• 
  A largely American bubble, for now

  Valuation data suggests that the froth around AI is far from evenly spread across global markets. In the United States, forward price-to-earnings ratios have risen sharply since end-2022, from about 17.4 to just over 22 by mid-2025, as AI expectations are priced aggressively into equities. Other developed markets have also seen some re-rating, but from much lower starting points. Forward multiples have increased in Germany, the UK, and Japan but remain well below US levels. Indian valuations, however, were already high before the AI surge and have expanded little, with some softening over the past year and a half as earnings growth caught up, and direct exposure to frontier AI remained limited. China and Hong Kong show similarly modest moves. This too marks a contrast with the dotcom era, when stretched valuations spread more widely across global markets rather than remaining anchored to the West
• Conclusion

  Globally, the speed and scale of capital flowing into AI leave little room for disappointment. Whether this marks the start of a lasting transformation or another bubble will depend on how quickly reality catches up with expectations already written into valuations.