📰 Key Takeaways

Moonshot AI (the company behind Kimi) is one of China’s most closely watched AI labs, and according to a Bloomberg report Friday, the company is targeting an annualized revenue run rate of $2 billion by the end of this year — double its August revenue run rate. This aggressive target reflects the strong growth of K3, the open-source model Moonshot launched this summer: even though K3’s usage has dipped slightly in recent months, OpenRouter platform data shows the K3 model family is still generating up to 300 billion tokens of usage per day. Still, Moonshot’s revenue estimates remain far behind OpenAI (reportedly around $40 billion recently) and Anthropic (around $65 billion). Since Moonshot’s model weights are fully open, its profit margins are much thinner than closed-source competitors — but the climbing revenue forecast shows that open-source AI models can still carve out substantial commercial territory, even with lower margins than closed frontier models. Meanwhile, Moonshot’s model development practices remain controversial, and possibly illegal: earlier this week Anthropic accused the company of engaging in long-running model distillation, routing nearly 300,000 Kimi requests directly to Claude Opus for processing — effectively having Opus pose as Kimi’s own model to answer users. Anthropic also said more than 23 million responses from its models were collected by Moonshot for training purposes.

💬 JudyAI Lab’s Take

Reports that Moonshot AI is planning to push its annualized revenue to $2 billion by year-end point to an expanding vision of what open-source model commercialization can look like — worth watching for anyone tracking the AI space.

Per the Bloomberg report, this target rests on the strong usage of the open-source K3 model — OpenRouter data shows the K3 family is still generating up to 300 billion tokens of usage per day, even with a slight dip in recent months. The takeaway for the AI builder community: even when the open-source path comes with far thinner margins than closed-source models (Moonshot’s revenue estimates still trail far behind OpenAI’s $40 billion and Anthropic’s $65 billion), sheer token volume alone can support a substantial commercial scale. At the same time, Anthropic’s accusation that Moonshot engaged in model distillation — routing nearly 300,000 Kimi requests to Claude Opus and collecting more than 23 million Claude responses for training — shows that behind the fast-growth open-source narrative, questions about data provenance and compliance remain very much unresolved.

For AI builders, when you’re evaluating the business case for any open-source model, it’s worth asking not just about the usage numbers, but whether its training data sources can hold up to scrutiny.

📅 Source Info

🔗 Further Reading