OpenAI was an American artificial intelligence research company and, during the 2020s, one of the defining institutions of the commercial AI sector. The company filed for Chapter 11 bankruptcy on 14 March 2031 in the Northern District of California, marking the terminal point of a three-year industry correction driven by the failure of the scaling hypothesis — the premise that adding more compute would reliably produce qualitatively new AI capabilities.
The company’s collapse was described by industry observers as a slow bleed rather than a crash: enterprise subscriptions declined over eighteen months, model releases landed without impact, and compute contracts were underwater by the time of filing. The bankruptcy represented the most visible institutional failure of the transformer-scaling paradigm, and the most consequential validation of the alternative approach pursued by competitors — particularly Anthropic and the Soviet RAZUM programme — who had diversified away from pure pretraining scaling years earlier.