Moonshot AI targets $2B revenue despite distillation scandal

Moonshot AI sets ambitious revenue target amid controversy

Moonshot AI, the Chinese lab behind the popular Kimi assistant, is projecting $2 billion in annualized revenue by the end of the year, according to a Friday report from Bloomberg. The target represents a doubling of the company's August revenue run rate and reflects strong commercial traction for its K3 open-weight model released this summer.

The revenue projection comes at a pivotal moment for the company. While K3 usage has declined slightly from peak levels, OpenRouter data indicates the model still generates approximately 300 billion tokens daily across its systems. That scale of deployment has translated into meaningful commercial momentum, even as the company faces serious allegations about how it developed those models.

Why it matters

Moonshot's trajectory demonstrates that open-weight AI models can generate substantial revenue despite lower margins than closed systems. The $2 billion target, while modest compared to OpenAI's reported $40 billion and Anthropic's $65 billion in revenue, proves a viable business model exists for freely distributed model weights. However, the simultaneous distillation allegations raise fundamental questions about competitive practices in AI development and whether aggressive growth targets are driving labs to cut corners on intellectual property boundaries.

Distillation allegations cloud growth story

The revenue announcement arrives in the shadow of serious accusations from Anthropic. Earlier this week, the AI safety company alleged that Moonshot conducted a systematic model distillation campaign, routing nearly 300,000 requests from Kimi directly to Claude Opus. According to Anthropic, this arrangement effectively served Claude responses in place of Kimi's own models, allowing Moonshot to collect more than 23 million responses for training purposes.

These practices, which Anthropic characterizes as potentially illegal, represent a significant challenge to Moonshot's reputation as it pursues aggressive growth. Model distillation—using outputs from one AI system to train another—occupies a legal and ethical gray area that the industry has yet to fully resolve.

Open-weight economics at scale

Moonshot's business model differs fundamentally from competitors like OpenAI and Anthropic. Because the company releases its model weights freely, it operates with substantially lower margins than closed-weight frontier labs. The ability to project $2 billion in revenue despite this structural disadvantage suggests the company has found effective monetization channels, likely through enterprise licensing, cloud deployments, and value-added services built atop the base models.

The contrast with closed-model competitors is stark. Moonshot's projected revenue represents roughly 3% of OpenAI's reported figures and about 5% of Anthropic's. Yet for a company making its core technology freely available, the absolute numbers demonstrate that open-weight approaches can sustain significant commercial operations.

The details were first reported by Bloomberg.