Pluralistic: How an AI moratorium can save AI bosses (16 Sep 2026)

Cory Doctorow argues that the “hyperscaler” AI model is fundamentally unprofitable due to massive expenditures and negative unit economics, with companies allegedly faking financials and using “novel, secret forms of mathematics” to claim profitability, akin to WeWork’s accounting fraud.
The AI industry is caught in a “Red Queen’s Race” because chatbots have extremely low switching costs, allowing customers to easily move between providers (e.g., OpenAI to Anthropic’s Claude). This prevents companies from establishing stable profits and forces continuous, costly innovation.
AI bosses are leveraging “criti-hype” around “superintelligence risk” to push for an AI moratorium. This would serve as a pretext for governments to grant them consent decrees, absolving them from antitrust laws and allowing them to collude to halt competition and ban open-source alternatives like Chinese “open weight” models.
Doctorow warns against granting AI companies antitrust immunity based on “superintelligence” fears, suggesting that their history of anti-competitive behavior and vast financial losses warrant government investigation for securities fraud, trafficking in child sex abuse material, and other potential crimes instead.