Moonshot AI, a leading Chinese artificial intelligence laboratory, is betting that its widely-used open-weight model can translate into substantial commercial success. According to a Friday report from Bloomberg, the lab has set its sights on reaching $2 billion in annualized revenue by year's end — a figure that would double its reported revenue run rate from August. This ambitious target underscores the momentum behind the company's K3 model since it launched in the summer.

EMBED_PLACEHOLDER_0

Although K3's usage numbers have dipped somewhat over recent months, data from OpenRouter indicates that K3 models are currently generating up to 300 billion tokens per day across the platform. Even so, Moonshot's revenue projections remain modest compared to industry giants OpenAI and Anthropic, which recent reports estimate at $40 billion and $65 billion respectively.

Since Moonshot makes its model weights openly available, the company operates with significantly thinner margins than competitors who keep their weights closed. Nevertheless, the growing revenue forecasts suggest that open-weight AI models can still generate meaningful income, even if they don't match the profitability of closed-weight frontier models.

That said, Moonshot's approach to model development continues to draw controversy — and potentially legal trouble. Earlier this week, Anthropic levied accusations that Moonshot conducted a prolonged model distillation operation, directing nearly 300,000 requests from Kimi straight to Claude Opus, essentially using Opus to serve responses in place of Kimi's own models. Anthropic further alleged that over 23 million responses were harvested from Anthropic's models to train Moonshot's systems.