The Dead Economy Theory: AI's Trillion-Dollar Bet Against Its Own Customers

The AI industry’s valuations only make sense if the product is large-scale labor replacement. OpenAI, Anthropic, and their peers have absorbed hundreds of billions in capital chasing a market large enough to justify the spend, and the only one that fits is the global labor market. Benchmarks like OpenAI’s GDPVal and the AI Productivity Index now explicitly measure model performance against professional occupations — bankers, consultants, lawyers, physicians — making the targeting reticle plain even as marketing softens it with words like ‘copilot’ and ‘augmentation.’

The consequence is a self-defeating loop. Firms that automate workers see immediate margin expansion and stock-price rewards — Block’s stock jumped 25 percent after Jack Dorsey cut nearly half his workforce citing AI coding agents. But displaced workers are also customers, and as spending contracts across the economy, the firms that fired them lose the demand that justified the cuts. Wharton economists Falk and Tsoukalas call this ‘The AI Layoff Trap’: in competitive markets each firm captures the full savings of automation but bears only a fraction of the demand destruction, creating a prisoner’s dilemma where everyone over-automates. CEO herd behavior accelerates the cuts beyond what efficiency would dictate.

The usual rebuttal — that economies have absorbed automation before — elides the timescales. Agricultural displacement took 140 years; the Industrial Revolution took 70 before wages recovered, with social upheaval in between. Former NEC official Bharat Ramamurti notes the AI shock could unfold in two years rather than the decades of the China shock, because investors need the model providers to generate revenue fast. Unlike narrow past automation, general-purpose AI threatens cognitive labor across every industry simultaneously, echoing Leontief’s 1983 warning that humans could become economically obsolete the way horses did.