Are We in an AI Bubble? Lessons from History’s Biggest Crashes
Exploring the parallels between the AI boom and the dot-com bubble, the risks of financial mania, and the lessons history teaches us about technological revolutions.
When the man leading the biggest AI wave on the planet says he is worried people are about to lose a huge amount of money, it’s time to pay attention. Sam Altman, the founder of OpenAI, has openly expressed concern about the scale of financial losses coming from the AI boom. Just days ago, the IMF and the Bank of England issued warnings about a bubble similar to the dot-com era, highlighting the risk of a sharp, sudden correction in global markets.
Some say we are in a bubble. Others argue there is no danger, and that this is simply what a technological revolution looks like. So, who is right? To find an answer, we need to understand both sides of the argument and look at what history says about big tech shifts and the financial manias that often accompany them.
The Skeptic’s Case: Why the AI Boom Looks Like a Bubble
Since ChatGPT captured public attention in late 2022, the value of U.S. stock markets has surged by trillions of dollars, with much of this growth concentrated in a handful of AI-related giants. This rapid ascent has created a palpable sense of unease, reflected in the skyrocketing number of Google searches for “AI bubble.”
So, what is fueling this fear?
1. The Circular Money Flow
A primary concern is the appearance of “circular money washing.” Take the relationship between big tech and AI startups. Cloud giants like Microsoft, Google, and Amazon invest billions into companies like OpenAI and Anthropic. In turn, these AI startups spend a huge portion of that capital on cloud services and chips provided by the very giants who invested in them.
The money moves in a circle: it leaves as an investment and returns as revenue. This creates the appearance of booming sales, fooling investors and inflating stock prices until the bubble eventually bursts. This pattern is hauntingly similar to the accounting tricks that fueled the dot-com bubble.
Back then, the magic word was “internet.” Companies with no product, profits, or business model saw their stock prices skyrocket. The growth was driven by infrastructure spending on fiber-optic cables. But behind the scenes, much of this growth was an illusion. For instance, telecom companies would swap unused fiber capacity with each other, booking the deals as billion-dollar revenue without any real money changing hands. When reality caught up, the collapse was catastrophic.
Many observers believe the same pattern is happening today with AI, but on a much larger scale.
2. Aggressive Accounting and Tiny Revenues
Beyond circular financing, companies are using accounting maneuvers to flatter their numbers. For example, some firms are recording expensive, short-lived AI chips on their books as if they last for many years. This spreads the massive cost over time, making annual expenses look artificially low and profits artificially high. Other companies have started shifting their AI expenses into separate external entities, hiding the true cost from investors.
These financial games are happening against a backdrop of enormous spending and minimal returns. The AI infrastructure industry is expected to reach $300 billion in spending this year, yet total AI revenue in the U.S. is only around $12 billion. An MIT study found that 95% of companies adopting AI have not yet generated any financial return. Even with a tool as popular as ChatGPT, less than 2% of its users are paying subscribers.
This massive gap between spending today and revenue tomorrow is what many analysts point to as the clearest proof that we are in a bubble.
The Bull’s Case: Why This Time Might Be Different
To be fair, there is a compelling counterargument. Many confident investors believe this is not madness, but rather the price of building the future. They argue that the current situation is fundamentally different from the dot-com era.
