The current AI boom dwarfs the 1990s tech bubble in scale, with US businesses investing nearly $1.5tn in IT equipment and software in 2025—more than triple the inflation-adjusted peak of the TMT era. Tech investments now drive 93% of US GDP growth, up from just 60% during the dot-com peak.

Hyperscalers including Microsoft, Alphabet, Amazon, Meta and Oracle plan capital expenditures growing 20% annually through 2030, while revenues are expected to rise only 15% per year. Under optimistic assumptions, only Amazon achieves positive returns on AI investments, with others facing negative ROI that could destroy trillions in shareholder value.

To generate a respectable 10% return, these companies would need to find an additional $2tn to $5tn in annual revenue—a near-impossible task for firms currently generating just $1.5tn combined. The math suggests a reckoning is inevitable.

Major AI IPOs from OpenAI and Anthropic may simply transfer risk to retail investors before reality sets in. While 2026 optimism could persist, the bubble likely bursts in 2027 or 2028, potentially triggering a market crash rivaling the early 2000s dot-com collapse.