Are there similarities between what we are seeing today with artificial intelligence, and the massive investment required to build the data-center infrastructure that supports it, and the dot-com collapse or the housing crisis that led to the Great Recession? It’s a reasonable question, and one worth examining.
To answer it, we need to go back to the late 1990s.
As the calendar approached the year 2000, companies poured billions of dollars into technology upgrades to prepare for Y2K. Dire predictions circulated about elevators failing and planes falling from the sky when the clock rolled over to January 1, 2000. None of that happened. But the investment surge was real.
At the same time, the internet was rapidly gaining traction. Businesses were building websites, consumers were beginning to shop online, and entirely new business models emerged that relied exclusively on the internet. Innovation was real, but so was speculation. Stock prices soared, especially in technology shares. The NASDAQ Composite nearly doubled between 1998 and its peak in early 2000.
When earnings and profits failed to match lofty expectations, valuations collapsed. Remember the infamous pets.com. The NASDAQ ultimately fell nearly 80% from peak to trough, contributing to the 2001 recession. It would take roughly 15 years for the index to fully recover the value lost during the dot-com implosion.
As always, investors then went searching for returns elsewhere. Capital flows to the highest rate of return, adjusting for risk.
In the mid-2000s, that search increasingly led to structured mortgage products, most notably mortgage-backed securities and collateralized debt obligations. These instruments pooled mortgage payments and passed the cash flows through to investors. Because housing prices had risen steadily for decades, these securities were widely viewed as lower risk.
Demand surged. To meet it, lenders originated more mortgages, often with weaker underwriting standards. Adjustable-rate mortgages proliferated, loan-to-value ratios climbed, and in some cases mortgages exceeded the value of the homes themselves. As long as home prices kept rising, the system appeared stable.
That stability proved illusory. When interest rates reset higher and borrowers began missing payments, the cash flows supporting these securities deteriorated. Losses spread quickly through the financial system, triggering the 2008 financial crisis and the Great Recession, the most severe economic contraction since the Great Depression. The pets.com implosion years earlier ultimately turned into the Great Recession.
So how does artificial intelligence fit into this historical comparison?
Once again, we are witnessing massive investment tied to transformative technology. Hundreds of billions of dollars—and potentially more than a trillion globally over the coming decade—are being invested in data centers, power infrastructure, and advanced semiconductor capacity to support AI. These investments are helping fuel equity markets, with a small group of large technology firms—the so-called “Magnificent Seven”—accounting for a disproportionate share of recent stock market gains.
As in prior cycles, leverage is playing a role. Much of this build-out is being financed with debt. While some of that debt is long-term, concerns are emerging about mismatches between financing structures and the underlying assets. Data centers may last decades, but the chips inside them often have useful lives measured in just a few years. Financing rapidly depreciating technology with long-dated debt introduces risk.
Markets have already shown sensitivity to that risk. Recently, disappointing news from a handful of AI-infrastructure firms triggered sharp reactions in equity prices. High expectations are embedded in today’s valuations, and much must go right for projected returns to materialize. Ultimately, someone must service the debt and deliver returns to capital providers.
That does not mean an AI-driven collapse is inevitable. There are important differences from past cycles. Many of today’s leading technology firms are profitable, cash-rich, and generating real revenue growth. AI is delivering tangible productivity gains, not just speculative promise.
Still, history offers a cautionary lesson. Periods of transformative innovation are often accompanied by overinvestment, financial excess, and unrealistic expectations. When returns fail to materialize as quickly or as broadly as hoped, markets adjust—sometimes abruptly.
The risk is not artificial intelligence itself. The risk lies in how aggressively it is being financed, how optimistic the assumptions have become, and whether capital discipline is maintained. History doesn’t repeat, but it often rhymes—and investors would be wise to remember that as the AI investment cycle continues to unfold.
Add a Comment