The Token Trap: Why AI Training Is the Only Way Out

The American economy has become the AI trade. In Q1 2026, GDP grew at 2% annualized, with AI-driven investment contributing about 75% of the increase. AI data centers, hardware, and networking hit 1.4% of US GDP, doubling from 0.7% and making AI infrastructure the leading driver of US private investment growth. Data from the St. Louis Fed shows AI investment accounted for 39% of marginal GDP growth over the trailing four quarters — bigger than the tech sector's 28% contribution at the peak of the dot-com boom. Excluding these investments, growth in the first half of 2025 would have been 0.1% annualized, near standstill. In 2026 alone, big tech's AI capital expenditure will pass $800 billion.

This infrastructure spend was justified by belief in AI's future, but is increasingly justified by revenue growth from the labs themselves. That's the contract: as long as token consumption keeps rising fast enough, the capital keeps flowing.

The Shift from Seats to Usage

Last year, the biggest Wall Street narrative was about an AI bubble. The underlying math was simple: $20 to $200 per month times the number of addressable knowledge worker seats could not justify trillions in infrastructure spending. The total addressable market of AI sold as seats wasn't going to cut it.