AI Boom Mirrors Dot-Com and Housing Bubbles — And May End the Same Way

A veteran of two major boom-bust cycles argues the AI investment frenzy shares the same structural DNA as the dot-com crash and the Great Financial Crisis: explosive real demand combined with dangerous leverage. The pattern is eerily familiar — genuine growth masking a financing edifice that eventually collapses under its own weight.

The demand side is undeniable. Anthropic's revenue hit $4.8 billion in Q1 2025, then more than doubled to $11.6 billion in Q2, putting it on a $65 billion annualized run rate — a 7x increase year-over-year that the author claims is unprecedented in corporate history. The company is now reportedly profitable, demolishing the prevailing narrative that frontier AI models can't make money. OpenAI, meanwhile, is doing roughly $40 billion in annualized revenue, though its sequential growth slowed sharply to just 18% in Q2 (from $5.7 billion to $6.7 billion) — framed here as the moment a racecar swerves off the track while a competitor blows past.

The leverage side is where the alarm bells ring loudest. Alphabet's off-balance-sheet commitments ballooned by nearly $500 billion in just three months — a figure that would take five full years to retire at the company's prior free cash flow rate of $100 billion annually. Notably, Alphabet burned cash for the first time in its public-company history in Q2. Meanwhile, Nvidia has structured $1.5 billion in investment plus $105 billion in credit toward an Ohio data center whose owner will likely use that money to buy Nvidia chips — and separately committed $30 billion to OpenAI, which will in turn lease compute. These circular financing arrangements are legal, but they function as vendor-financed leverage that artificially amplifies demand across the entire AI ecosystem.

The historical parallel is stark: the internet boom saw hundreds of billions borrowed to build out telecom networks; the housing boom saw lending standards progressively gutted until there was no one left to lend to. In both cases, the music stopped when supply caught up with demand and leverage hit its ceiling. A rising interest-rate environment or a sudden loss of investor confidence could serve as the trigger this time — though the author is candid that even expert investors like Stan Druckenmiller held on too long in 2000, knowing full well the dot-com boom was a bubble.

The author's bottom line isn't panic — it's calibrated exposure. AI represents one of the largest economic opportunities in history and a bubble simultaneously, and those aren't mutually exclusive. The playbook: participate, learn, and exploit the technology's real benefits, but never stake more than you can afford to lose entirely. At Anthropic's current trajectory, a $1 trillion valuation may look reasonable given the $1.4 trillion global enterprise software market — but sustaining hypergrowth will require cracking entirely new market segments, and history suggests that's where the stumbles begin.