“Looking at the implementation of enterprise AI through four major models over a week, why 95% of the pilot projects fail to make money”

In September 2026, four leading AI laboratories—Anthropic, Google, Meta, and OpenAI—announced flagship AI models within the same week. Anthropic launched Claude Fable 5.1 on September 1st, designed for long-term agent tasks, while Google released the long-term version of Gemini 3.8 Flash on September 2nd. Against the backdrop of frequent model releases, enterprise-level AI implementation faced severe challenges; data showed that 95% of pilot projects failed to achieve profitability. This phenomenon was referred to by CNBC as ‘model fatigue’, indicating that although models were updated frequently, the related funding and profits mainly went towards infrastructure rather than directly translating into commercial value.