The AI Bubble Could Burst Beyond Wall Street

For millennials like me, if there is one lesson that the 2008 crisis taught, it is that highly speculative financial dynamics have real-life consequences. Entire economies go into recession, jobs are lost, fiscal policies are tightened, and in the end, the everyday lives of citizens are affected. 

When John Maynard Keynes was talking about ‘animal spirits’ to describe the human emotions that drive economic decisions, he probably couldn’t fathom the trillion-dollar boom in AI investments that we are witnessing today. And the present’s animal spirits might be exorcised by reality, with repercussions felt on the global economy. 

At the end of October, legendary hedge-fund manager Ray Dalio – founder of Bridgewater Associates – spoke about the risk of an AI bubble in a market where 80% of earnings are concentrated within Big Tech firms ¹.

But his concerns are shared among many in the financial industry, including the Bank of England ². In fact, market concentration is getting more extreme: the “Magnificent Seven” tech firms are making up more than one-third of the US-index S&P 500, with massive valuations that are inflated by ambitious bets on future demand ³. 

The market concentration is also contributing to reinforcing inequality, not just between the Global North and the Global South, but also within the same Western societies ⁴. When private companies become this big, they can sit at the negotiating tables and exert influence over international politics.

In fact, a second factor heavily pushing investments is the geopolitical competition between the US and China. Considering that Nvidia’s value of over $5T is worth more than any GDP besides the two global superpowers, and the leveraging role that the AI chipmaker has for US international strategy, a burst in the AI industry has the potential to shake the world’s economy.

Understanding Circular Financing In AI

One of the most worrying signs of a potential AI bubble is the number of circular deals between the major firms of the industry, which alarmingly reminds of the capital arrangements that characterized the dotcom bubble at the end of the 1990s.

Taking a look at how OpenAI is financing its growth plans allows to have a better understanding of the circularity of capital in the Artificial Intelligence sector. Between 2019 and 2023, Microsoft invested more than $13 billion in OpenAI, with the latter gradually spending the same amount by buying cloud computing power from Microsoft ⁵. 

In a similar way, Oracle agreed to spend $300 billion on building new data centers that OpenAI is going to use for developing its technologies. OpenAI will then pay Oracle back in order to use these same data centers.

Last October, OpenAI announced a deal with chipmaker AMD worth tens of billions of dollars, in addition to a warrant that could allow the creators of ChatGPT to own around 10 percent of AMD in shares if certain conditions are realized ⁶. 

Roughly one month before, OpenAI reached an agreement with AMD’s competitor Nvidia, which is going to invest $100bn in the company. The investments will circulate back with OpenAI placing orders over time for a total capacity of 10GW of Nvidia chips.

Nvidia itself has used a similar financing strategy by investing in dozens of Artificial Intelligence startups that rely on Nvidia chips to develop their models ⁷.

Fragilities In The Market

There are other revealing signs of a potential AI bubble, in particular when looking at the infrastructure side of Artificial Intelligence. Data center operators bought GPUs to train AI models at peak prices (a new single processing unit cost $50000 in 2024), with the assumption of massive future demand. The same hardware can now be rented for just $2.80 per hour ⁸. 

This fall in prices, which clearly shows an oversupply situation, is interestingly only affecting smaller GPU-rental startups, which are now desperately cutting prices to stay afloat. While most buyers are sticking with ready-made AI services from companies like OpenAI and Anthropic rather than renting hardware, the demand that’s left comes from low-budget users. A large number of data centers and AI startups are likely to never recover their initial investments, with a wave of failures that is going to leave out only the bigger players.

Another striking example comes from the initial public offering of AI cloud computing company CoreWeave, which slashed its IPO size from an overestimate of $2.7 billion to $1.5 billion and brought in Nvidia as a major investor ⁹. At the same time, Nvidia constitutes the sole GPU supplier of CoreWeave.

There is an additional key factor at play: energy consumption. Just in the United States, data centers require a total capacity of 51GW, which is projected to need a further 44GW by 2028 ¹⁰. But the current energy infrastructure is able to cover only 25 additional GW. To have a clearer idea of these massive proportions, 1GW is equivalent to the output of a nuclear reactor.

Data centers have real environmental impacts on the communities around their location: residents of rural Georgia are complaining of decreasing water quality and access after Meta built a massive data centre nearby ¹¹. A 1MW data centre can use in one year the equivalent of the daily water consumption of 300.000 people ¹². It is clear how the expansion of the AI infrastructure can put additional stress, especially on regions already vulnerable to water shortages.

The Wider Risks Of the AI Bubble

This growing set of interdependencies between AI companies is clearly inflating the level of demand. Adding this to the concentration of the market, a setback of one of the bigger players has the potential to shake the whole industry ¹³.

But the risks go beyond the stock market, and a correction in global equities could affect investment and pension funds, retail portfolios, and even banks ¹⁴. Speculation on future demand, geopolitical competition, and overshooting capitalism are therefore creating the conditions of an AI bubble whose bursting could be felt across different economic sectors, ultimately affecting individual savers.

American investor and hedge fund manager Michael Burry has recently drawn a parallel with the dotcom bubble, highlighting how nowadays many regular investors have their savings in index funds and ETFs that are more exposed to AI firms ¹⁵. 

To be fair, several analysts are indeed more cautious in making comparisons with the dot-com bubble, especially when considering that the companies driving the AI rally are actually making massive amounts of money ¹⁶.

Either way, the aforementioned risks will be mostly felt in the Global North because the current economic benefits of AI innovation are flowing towards companies in Silicon Valley and their shareholders.

But for the rest of the world, AI firms are already continuing the extraction economy that disproportionately affects countries in the Global South by relying on low-paid workers to train their systems, while increasingly automating and making these same jobs replaceable ¹⁷. 

So, is it still possible to prevent the AI investment fever from further derailing?
On a regulatory level, AI innovation can not come without strong governance that safeguards society and maintains accountability through a rigorous risk management process ¹⁸. 

But if the finance sector has raised a lot of skepticism in its capacity to de-incentivise speculative dynamics, partnerships between countries and grassroots organizations have higher chances to create a tangible impact. In the US, grassroots opposition to data center development managed to block or delay $98 billion in projects just between April and June 2025 ¹⁹. 

For countries in the Global South, cooperation for building a model of data and labour sovereignty can help in protecting their societies from extractive economic dynamics ²⁰.

Overall, the risk of an AI bubble is accelerating awareness of the distortions of speculative finance and infinite-growth capitalism. It could be the wake-up call for a wider targeted action towards a more inclusive and democratized digital future.

References

1. Harring, A. (2025, October 28). Ray Dalio says a risky AI market bubble is forming, but may not pop until the Fed tightens. CNBC. https://www.cnbc.com/2025/10/28/ray-dalio-bubble-ai-federal-reserve.html
2. Mitchell, A. (2025, October 8). Bank of England warns of AI bubble risk. BBC. https://www.bbc.com/news/articles/cx2e0y3913jo
3. Carvão, P. (2025, October 16). Is AI a boom or a bubble? Harvard Business Review. https://hbr.org/2025/10/is-ai-a-boom-or-a-bubble
4. Anilkumar, A. (2025, May 13). AI inequality: When intelligence isn’t shared. Digital Peace. https://digital-peace.org/ai-inequality-intelligence-gap/
5. Metz, C., & Hirsch, L. (2025, October 31). How OpenAI uses complex and circular deals to fuel its multibillion‑dollar rise [Interactive]. The New York Times. https://www.nytimes.com/interactive/2025/10/31/technology/openai-fundraising-deals.html
6. Hammond, G. (2025, October 6). OpenAI targets 10% AMD stake via multibillion‑dollar chip deal. Financial Times. https://www.ft.com/content/bfafd06e-0a92-4add-9ae5-622e3c2c8f29
7. Forgash, E., & Ghosh, A. (2025, October 7). OpenAI’s Nvidia, AMD deals boost $1 trillion AI boom with circular deals. Bloomberg. https://www.bloomberg.com/news/features/2025-10-07/openai-s-nvidia-amd-deals-boost-1-trillion-ai-boom-with-circular-deals
8. Elder, B. (2025, October 16). What GPU pricing can tell us about how the AI bubble will pop. Financial Times. https://www.ft.com/content/d49707ae-5d6b-473e-9e2b-487d318e6fe9
9. Basu, T. (2025, October 27). The year’s biggest AI float just became an IPOuroboros. Financial Times. https://www.ft.com/content/daaf2a6a-27a9-46cd-b325-6ed497648f17
10. Smyth, J. (2025, December 8). The power crunch threatening America’s AI ambitions [Interactive]. Financial Times. https://ig.ft.com/ai-power/
11. Fleury, M., & Jimenez, N. (2025, July 10). ‘I can’t drink the water’ – life next to a US data centre. BBC News. https://www.bbc.com/news/articles/cy8gy7lv448o
12. Spindler, W., Hahn‑Petersen, L. A., & Hosseini, S. (2024, November). Circular water solutions key to sustainable data centres. World Economic Forum. https://www.weforum.org/stories/2024/11/circular-water-solutions-sustainable-data-centres/
13. Waters, R. (2025, October 12). How OpenAI put itself at the centre of a $1tn network of deals. Financial Times. https://www.ft.com/content/4e39d081-ab26-4bc2-9c4c-256d766f28e2
14. Financial Times. (2025, October 13). Measuring risk in the AI financing boom. Financial Times. https://www.ft.com/content/50f6a373-f7b9-455e-b8ab-129d312822c1
15. Harring, A. (2025, December 3). Why Michael Burry thinks the AI bubble will unravel. CNBC. https://www.cnbc.com/2025/12/03/why-michael-burry-thinks-the-ai-bubble-will-unravel.html
16. Yaffe‑Bellany, D., & Scheiber, N. (2025, December 9). Wall Street Is Shaking Off Fears of an A.I. Bubble. For Now.The New York Times. https://www.nytimes.com/2025/12/09/business/wall-street-valuation-ai-bubble.html
17. Apeagyei, K., & Murthy, A. (2025, November 14). The perilous future of AI work in the Global South. Media@LSE (London School of Economics and Political Science). https://blogs.lse.ac.uk/medialse/2025/11/14/the-perilous-future-of-ai-work-in-the-global-south/
18. Carvão, P., Ancheva, S., Atir, Y., Jeloka, S., & Zhou, B. (2025, February). Governance at a crossroads: Artificial intelligence and the future of innovation in America (M‑RCBG Associate Working Paper No. 251). Mossavar‑Rahmani Center for Business & Government, Harvard Kennedy School. https://www.hks.harvard.edu/centers/mrcbg/publications/awp/awp251
19. Data Center Watch. (2025). Q2 2025 update: 125 % surge in data center opposition. Data Center Watch. https://www.datacenterwatch.org/q22025/
20. Maughan, K. (2025, August 3). What will the AI revolution mean for the global south? The Guardian. https://www.theguardian.com/commentisfree/ng-interactive/2025/aug/03/artificial-intelligence-global-south