Moonshot AI Targets $2 Billion Annual Revenue by Year-End

Revenue Target and Model Success

Moonshot AI, a prominent Chinese AI lab, has set an ambitious target of $2 billion in annualized revenue by the end of the year. This figure represents a doubling of the company's reported revenue run rate from August. The company attributes this aggressive goal to the success of its K3 model since its release this summer.

Open-Weight Model Performance

Despite a slight decline in usage figures in recent months, OpenRouter data indicates that K3 models are generating as many as 300 billion tokens daily on the system. While Moonshot's projected revenue is significantly lower than that of OpenAI and Anthropic, which report $40 billion and $65 billion respectively, the company's open-weight model approach yields lower margins. However, these rising projections suggest that open-weight AI models can still generate substantial revenue.

Controversy Over Model Development

Moonshot AI's model development practices have drawn controversy. Anthropic recently accused the company of a prolonged model distillation campaign. Anthropic alleged that nearly 300,000 requests from Kimi were routed directly to Claude Opus, effectively serving Opus in place of Kimi's own models. This process reportedly collected over 23 million responses from Anthropic models for use in Moonshot’s training.