Are We in an AI Bubble? Maybe. But That’s Not the Whole Story

​

There are credible reasons and warnings from investment banks, central banks, and other policy institutions that suggest there might be an alternative intelligence (“AI”) pricing bubble. For example, The Bank of England has noted that equity valuations for technology companies deeply involved in AI look “stretched,” particularly by traditional metrics. Likewise, the market surge in this field is in fact about mega-cap tech firms “Magnificent Seven”, which already have strong business models and financials and already capture large proportions of market return or capitalization, aim to open new frontiers with AI expansion. Similar warnings came from the European Central Bank. The IMF has also cautioned about AI-driven tech stock overvaluation and that markets may be overly optimistic. Thus, if investor sentiment shifts, a correction in one of those mega-cap tech firms may have contamination effect.

​

Of course there are counter arguments too. Some investment banks still believe that while valuations are high, they are supported or justified by real growth and demand (especially for infrastructure, semiconductors, AI compute, etc.). For example, Goldman Sachs has at times pushed back on the bubble narrative, saying valuations may be strained but not necessarily in a crash territory yet. According to others, much of the risk depends on whether AI’s expected productivity gains, regulatory environment, supply chains, etc., deliver. If they do, some of the concerns may look overcautious in hindsight.

​

There are significant reasons to believe that burst of the AI bubble may be waiting around the corner. Because I think this discussion will boil down to market regulation policy discussion in the end, not to engaging in the discussions but organizing my mindset seems to be the best way to proceed.

​

The financial bubbles may be traced all the way back to humankind’s curiosity about what is waiting in the future. Robust foresight of the future will provide some sort of advantages for early movers in financial markets, in some cases the gains for the early movers and risk takers may be very ve lucrative indeed.

​

But particularly when it comes to appreciation of promising new businesses, valuation models, tend to use assumptions that are heavily dependent on optimistic growth, discount rates etc. This flawless execution and uninterrupted growth perspective may bring unsustainable growth destined for collapse which refers to three linked things: valuation dynamics[1], capital flow behavior[2], and psychology[3] have entered a phase that can no longer be supported by the fundamentals of the underlying business or technology[4].

​

Everyone and everything are desperately in need of external proof of recognition and affirmation, so as valuations. If it cannot bring an external proof, valuations disappear, so as the over excitement depending on them. If the underlying assumptions like rapid adoption, large productivity gains, minimal disruption, no significant technical or regulatory setbacks prove too optimistic (e.g. slower adoption, regulatory hurdles, supply chain problems, energy / data bottlenecks, consider the recent slowdown in the shift to electric vehicles), there is risk of disappointment. is the classical sense we understand a bubble.

​

This perspective, on the other hand, provokes insatiability and easily manipulable nature of humankind as if the outcomes of these models have been materialized. Having stated this, I do not aim to be more poetic than analytical, but it is obvious that not everything flows according to a ratio in financial markets because no matter how much algorithmic trading there is, what happens in financial markets reflects human nature.

​

No one can correctly decide when or after what a bubble may burst. We recall it when we see a burst[5]. Although an ending that we may call “burst” is not inevitable, there may be sharp corrections etc., from the beginning we know that nearly all bubbles end with a burst or significant correction rather than quietly deflating. Because this burst or sharp correction brings real economic consequences, several precautions may well be considered by the regulators to prevent systemic risk and let the risk takers be aware that they are taking risks.

​

I am not against bubbles in the markets, and I do not think this is something from which we should refrain. Obviously, the dynamic that causes financial bubbles is somewhat the same dynamic supporting financial innovation. And I at the same time observe that prices rationalize after a collapse, but the experience and sometimes the technology developed does not fade away. The bubbles in the markets therefore are inevitable and acceptable, as long as they do not cause serious risk to the financial system.

​

In this regard, recall the sweet downhill path provided to the market players and the catastrophic magnifying effect of excessive misuse/abuse of derivative instruments in collaboration of the parties engaged during the credit crunch occurred in the first decade of the century. This AI bubble, if there is so, is a bit different though. Because here the value created does not solely depend on mathematical models and producing new derivative instruments on already derivative instruments. Here valuations may be deemed to be flawed with their excessively optimistic assumptions and market pricing mechanism may be deemed to be flawed with its view of continuous exponential demand adding an extra dimension to it.

​

The Bank of International Settlement brought its own perspective and has discussed the implications of AI for the economy and central banks—including that AI could alter financial stability, while many AI systems are opaque, and their effect on inflation, labor, productivity is still not fully understood. While not explicitly calling it a "bubble" in all cases, BIS is warning that central banks must “raise their game” in assessing risks. Strategists at Bank of America have stated that certain price actions, concentration, valuation, and speculation all "appear frothy (over excited, over hyped)" around AI, though he also notes central banks have not yet responded by tightening explicitly for AI.

​

AI Therefore, proposing fine prudential interference with excessive use of synthetic financial instruments tied to model performance of AI business would be meaningful in this regard.

Accompanying this, valuation models, in addition to the current ideal execution assumptions, may be required to assume tempered execution, benchmarked to historical technology adoption or return on investment patterns.

​

These two, accompanied with a “caveat emptor” requirement about dependence of asset valuation reports on sometimes over optimistic assumptions may be considered as a response to bubble formation concerns without disregarding human nature.

Public companies that deploy or sell AI often include explicit risk-factor language that acknowledges market volatility and “bubble” risk along with warnings for investors about technological, regulatory and IP uncertainties where relevant, such as:

• Meta Platforms (10-K) — includes typical forward-looking risk-factor language about a fast-changing environment and uncertainty around technologies and regulation. (See Meta 10-K “Risk Factors.”)

• Amazon (10-K) — mentions risks from AI and other technological changes in its risk disclosures.

• SES AI / other AI/tech firms (10-K) — explicit language about AI algorithm flaws, regulatory uncertainty, cybersecurity exposure, and that “there can be no assurance that the usage of such technologies will enhance our products or services” — a concrete statement companies use to warn investors about AI performance and valuation risk.

 

Filings explicitly using the phrase “bubble market” — some companies include a risk-factor sentence warning that the market has experienced “so-called ‘bubble market’” behavior in technology sectors, noting that such bubbles can significantly affect trading prices.

 

[1] Prices of assets (e.g. stocks, funds, derivatives) rise much faster than earnings, revenues, or measurable productivity gains. Traditional metrics like price-to-earnings (P/E) or price-to-sales (P/S) ratios reach levels that can only be justified if extraordinary growth continues indefinitely. If earnings cannot cathc up expectations should fall.

[2] Money flows into the sector faster than it can be productively absorbed. Companies start raising capital not to meet genuine demand but because investors are willing to fund anything associated with the fashionable theme

[3] Because bubbles feed on powerful narratives — “this time is different,” “AI will change everything,” “you can’t afford to miss out.” this language override critical analysis and create a reflexive loop: rising prices validate the story, which attracts more buyers, pushing prices even higher

[4]During the Japanese real estate bubble in 1980s, it is said, ‘the valuation of the Imperial Palace grounds in Tokyo was greater than that of all the real estate in California combined when the bubble was at its peak in 1989’. In 1991, the bubble went bust, eventually stagnating the Japanese economy for almost a decade. (https://www.equentis.com/blog/lessons-learnt-from-the-7-famous-investing-bubbles-in-history)

[5] The break of the spell is often triggered by disappointing earnings, regulatory shifts, or macro shocks, early investors start cashing out, growth slows, liquidity reverses drastically correct.

​

(This section was prepared with the assistance of OpenAI’s ChatGPT (GPT-5 model), which was used to summarize and synthesize information from publicly available sources identified through web searches (October 2025)).