Analyzing the current dynamics of the markets, I see parallels between the artificial intelligence boom and the infamous dot-com crash of the late 1990s. However, as fresh data shows, the scale of the current AI bubble is not just comparable—it has surpassed its predecessor across a number of critical metrics.
The key difference lies in the volume of investments. While total infrastructure spending during the dot-com era was estimated at around $500 billion, in the AI era this figure has soared to an astronomical $5.8 trillion. This is not just growth—it is a tectonic shift in the scale of capital investment.
A similar picture is seen in the IPO market. The largest IPO during the dot-com era raised about $4 billion. Today's record holder—$86 billion. That is a 21.5-fold difference. This indicates an unprecedented appetite among investors for companies that often have not yet generated a single cent of profit.
Particularly alarming is the valuation of unprofitable companies. In the dot-com era, the most expensive of them was worth about $100 billion. Today, its counterpart is valued at $965 billion—almost ten times more. The market is literally churning out "unicorns" from thin air, ignoring fundamental metrics.
Signs of Overheating and Systemic Risks
The signals of overheating are obvious. Companies go public at the peak of hype and then rapidly lose market capitalization. For example, SpaceX went negative within a month after the largest IPO in history. This is a classic bubble pattern.
Another alarming sign is buyer financing schemes. Nvidia, similar to Lucent and Nortel in the late 1990s, has launched a program to finance its own customers. Let me remind you that it was precisely this behavior that led to some of the largest bankruptcies in U.S. history.
Why the Current Bubble is More Dangerous
The main threat, in my view, lies in the fact that the dot-com bubble was primarily a stock market bubble. The current one, however, includes private credit, project bonds, and insurance company money. All these funds are flowing into data center infrastructure, creating a multi-layered structure that, upon collapse, could trigger a cascading effect.
Examples already exist. Blue Owl froze investor withdrawals while simultaneously financing 80% of Meta's $250 billion campus in Louisiana. Oracle's rating was downgraded to nearly "junk" status: half of its order book is tied to OpenAI. Anthropic has $90 billion in capacity lease obligations with zero profit before its IPO.
My conclusion: AI technology is real and will change the world, just as the internet once did. But the question is not about the technology, but about the price. Do the companies justify the sums being paid for them? Public reports in the coming quarters will provide the answer. The market is pricing in future super-profits, but history teaches that between a "real technology" and a "bubble" lies a vast chasm, and most companies from the dot-com era did not cross it.