A veteran of the late-1990s internet infrastructure boom warns that today’s artificial intelligence investment surge mirrors the speculative excesses of the dot-com era. Both periods promised sweeping transformations—one through digital commerce, the other through AI-driven automation—yet both ignored core financial principles for extended periods. The parallels extend beyond hype to structural vulnerabilities in how these industries finance growth.
The late-1990s bubble thrived on the assumption that conventional financial metrics no longer applied. Infrastructure providers selling high-speed networking equipment were valued as if their success was preordained, even though their business models depended on unsustainable debt loads. An engineer who worked for one such firm recalled a presentation featuring the Sierra Nevada mountains and the slogan: “In 1848, you didn’t want to be panning for gold. You wanted to be selling shovels.” The implication was that infrastructure providers would prosper regardless of whether the companies using their products succeeded. The company justified valuations by arguing that earnings before interest, taxes, depreciation, and amortization (EBITDA) no longer mattered. Reality proved otherwise.
By 2001, the telecom infrastructure firm where the writer worked filed for bankruptcy, joining others like Global Crossing, WorldCom, and Nortel Networks. Nortel’s stock value plummeted by roughly 99%. Many other infrastructure providers lost far more: Global Crossing and WorldCom filed for Chapter 11 bankruptcy, while Lucent also saw its stock value collapse. Most failures stemmed from aggressive expansion financed through debt, often arranged in circular credit schemes where companies extended loans to their own customers. When the market corrected, these debts became uncollectable, triggering a cascade of collapses. The first big sign that the sector was going under in 2001 was American Express’s decision to suspend the writer’s company credit cards, which came some months before the bankruptcy filing.
Read Also: Heritage Energy ramps up Uzere oil spill cleanup
Similar financing patterns are now emerging in AI. Anthropic, for instance, depends on computing capacity from Nvidia, which is also one of its largest investors. Microsoft has invested over $13 billion in OpenAI while generating revenue through Azure cloud services, which hosts OpenAI’s operations. These interlocking relationships create a fragile ecosystem: if demand for AI services weakens, the entire chain could collapse. During the dot-com crash, the first red flag appeared when American Express cut credit lines to the writer’s company months before its bankruptcy. Today, the risk is that financiers may abruptly withdraw funding from AI firms if returns fail to materialize, leaving trillions in debt exposure vulnerable.
One critical difference lies in scale. In 2001, the debt involved was substantial but contained. Now, AI financing resembles the pre-2008 financial crisis in both magnitude and concentration. The technology, media, and telecommunications sector now represents nearly half of the U.S. large-cap equity market, up from one-fifth during the 2008 financial crisis. If AI companies cannot generate sufficient revenue to service their debt, the economic fallout could surpass any previous speculative bubble.
Removing trillions from Western payrolls would destabilize economies, potentially triggering a financial crisis. Alternatively, if markets conclude that AI firms cannot repay debts, the crash could occur first, followed by broader societal disruption.
