When AI Costs More Than the Workers It Replaced: Inside the Token Price Spike

Since ChatGPT's release, AI appeared capable of anything -- hyperrealistic images, code generation, delegating complex tasks -- and companies rushed to adopt it once they saw how cheap it was. But that cheapness soon created problems of its own.

As Anthropic raised alarms about AI's impact on employment, tens of thousands of layoffs hit tech companies including Amazon, Chegg, Microsoft, Meta, and Salesforce. Fast-food chains started replacing drive-thru staff with AI too -- it looked like AI was wiping out jobs across every industry.

But a problem has since surfaced. According to Bloomberg, nearly half of all US data-center construction projects planned for 2026 have been canceled or delayed due to component shortages. Meanwhile, demand for AI infrastructure is reported to still exceed supply -- but experts are starting to question how much of that demand is real usage versus inflated figures.

Component shortages, delayed builds, and surging global adoption combined to drive token costs up sharply. Concerns over rising AI costs contributed to a major tech stock sell-off that dragged the S&P and Nasdaq down more than 500 points.

Uber and Microsoft are starting to second-guess aggressive AI adoption, as AI operating costs begin exceeding the cost of human employees in some scenarios. According to Reuters, running AI agents today already costs about the same as human labor -- and sometimes more. AI remains clearly cheaper for coding tasks, the gap is minimal for data entry, but in areas like call centers, human labor is now actually the cheaper option.

AI companies have been able to subsidize the real cost of tokens through abundant funding so far, but that appears to be ending. With Anthropic's and OpenAI's public listings, investor pressure for returns, and ongoing component shortages, the conditions are in place for token costs to keep rising.