AI Bubble Threatens Economic Stability Amid Regulatory Lapses

As I walk through the Financial District, the chill in the air feels different this year. It’s not just the coming winter but a familiar, metallic scent of speculative fever. I’ve covered market manias before—the dot-com frenzy, the housing bubble, the crypto craze. Each had its own signature, its own internal logic that made sense until, suddenly, it didn’t. What I’m seeing now, in the earnings reports, the debt issuances, and the whispered conversations after investor calls, is something of a different scale. It’s the AI boom, and a new report from policy experts sounds an alarm we would be foolish to ignore: this bubble could burst and it could take the broader economy with it.

The analysis, penned by Matt Scherer of the Open Markets Institute and Maya Jenkins of Americans for Financial Reform, frames a staggering reality. The U.S. stock market’s valuation is now inextricably linked to artificial intelligence. The nine most valuable American companies are all tech firms betting their futures on it. This isn’t niche investing; it’s the core of our market. And the spending is breathtaking—a $7 trillion sprint to build data centers and infrastructure for generative AI models. Early on, this was funded by the immense profits of Big Tech. But that well is running dry. The costs are exploding while the revenues from actual AI products remain, in the grand scheme, meager. So these companies are turning to debt and lots of it.

Here’s where the numbers get concerning. Nikkei Asia estimates that just five tech giants have accrued roughly $3 trillion in debt. More unsettling is the detail that $1.65 trillion of that is reportedly hidden off their formal balance sheets, using financial arrangements that echo the shadowy tactics of Enron. To offer some historical perspective, that total debt figure now surpasses the size of the entire subprime mortgage market at its 2007 peak. We all remember what happened next.

The fundamental math is becoming a serious problem. The report’s authors calculate that to service these mounting bills, AI companies would need to generate an unprecedented $2 trillion in new annual revenue. Given the current landscape—frenzied competition, soaring infrastructure and energy costs, and persistent reports from businesses seeing little return on their AI investments—that target looks increasingly fantastical. The gap between spending and income isn’t closing; it’s widening. This turns the question from if a correction comes to when.

• Market valuation increasingly tied to AI
• $7 trillion spent on AI infrastructure
• $3 trillion in debt among tech giants
• $1.65 trillion hidden off balance sheets
• Need for $2 trillion in new annual revenue
• Regulatory risks and lack of oversight

A common retort I hear on the trading floors is, “So what? It’ll be like the dot-com bubble. The weak startups will fail but the giants will endure.” This is a dangerous miscalculation. The giants of 2000, like Cisco or Intel, were not leveraged to the hilt the way today’s behemoths are. More critically, today’s ecosystem is a dense web of circular financing. These tech titans and the AI startups they fund are deeply intertwined in a complex lattice of investment and debt, creating a systemic risk that didn’t exist 25 years ago. One firm’s stumble could trigger a cascade.

Worse, this risk is already leaking into the wider financial system, much like subprime mortgages were sliced, diced, and sold as safe assets before 2008. Risky debt linked to data center projects is being repackaged and sold to institutional investors, including insurance companies. As The Financial Times has noted, private equity firms have been purchasing life insurers and loading their portfolios with higher-yielding, riskier debt. This is a direct channel for financial contagion. A shock in Silicon Valley could rapidly flow to Main Street retirees and policyholders.

In the face of these mounting risks, regulatory momentum is moving in the wrong direction. The Federal Reserve is moving to loosen bank capital requirements. Stress tests, designed to see if banks can withstand a downturn, are being weakened. Simultaneously, there are moves to funnel workers’ retirement savings into more opaque corners of the shadow banking system. These actions, as outlined in the report, reduce the financial system’s overall resilience and shift the risks of catastrophic loss onto the public. They make reckless speculation more attractive precisely when we need sober guardrails.

The path forward requires clarity and courage. Regulators, from the Fed to the Financial Stability Oversight Council, must use their existing tools. They should demand transparency on major institutions’ exposure to AI-linked debt and scrutinize adjacent risk pools like the $3 trillion private credit market. They must examine how bank balance sheets would withstand a sudden plunge in tech valuations. Crucially, as the authors argue, policymakers must commit now to a principle of “no AI bailouts.” The moral hazard of using public funds to rescue corporations from a crisis of their own making would be an outrage, undermining economic justice and public trust.

Bubbles operate on a logic of their own, divorced from fundamentals. There’s no telling how high this one can float or how long it can last. But the certainty is that markets won’t self-correct in time. The responsibility falls to regulators and lawmakers to wake up. They must safeguard the real economy—the jobs, savings and futures of working people—from the excessive risks being concentrated in a handful of tech stocks and their labyrinthine debt. The warning is on the tape. The question is whether anyone in power is willing to read it.