Moonshot AI Eyes $2 Billion Revenue Target as K3 Model Gains Traction

Moonshot AI, a leading artificial intelligence research organization based in China, is pursuing an ambitious financial milestone. According to a Bloomberg report released Friday, the firm intends to reach $2 billion in annualized revenue by year's end—a figure that represents a doubling of its August revenue run rate. The company's K3 model, launched this summer, has driven this optimistic projection.

Despite a modest decline in K3 usage over recent months, the model continues to demonstrate substantial demand. Data from OpenRouter indicates that K3 models generate approximately 300 billion tokens daily on the platform.

Moonshot's revenue targets remain considerably smaller than those of its major competitors. OpenAI and Anthropic command projected revenues of $40 billion and $65 billion respectively, according to recent industry reports. However, because Moonshot distributes its model weights openly rather than keeping them proprietary, the company operates with significantly reduced profit margins compared to closed-weight rivals. The company's growth trajectory suggests that open-weight AI models can still generate meaningful revenue, even if they lack the financial upside of closed-weight frontier systems.

Moonshot's path forward remains complicated by questions surrounding its model development methodology. This week, Anthropic leveled accusations that Moonshot has conducted an extended model distillation operation, directing roughly 300,000 requests from its Kimi product directly to Claude Opus and presenting Opus responses as Kimi's own output. Anthropic stated that Moonshot collected more than 23 million responses from Anthropic's models for incorporation into its training process.