Kimi-maker Moonshot AI targets $2B in annual revenue
Chinese AI lab Moonshot AI is aiming for roughly $2 billion in annualized revenue by the end of the year, a figure Bloomberg says would be about double the company’s revenue run rate in August. The target follows strong market uptake of its open-weight K3 model after its release this summer, even as some usage metrics have softened recently.
Why It Matters
The plan highlights that open-weight foundation models can still generate substantial commercial returns despite lower margins than closed-weight rivals, and it comes amid allegations that Moonshot used large-scale data extraction from a competitor — a charge that raises legal and ethical questions about model training practices.
Key Facts
- Company: Moonshot AI
- Revenue target: $2 billion annualized by year-end (reported by Bloomberg)
- August run rate: Target would be double Moonshot's reported revenue run rate for August
- Model: K3 (open-weight), released this summer
- K3 usage: Usage declined slightly in recent months, per report
Moonshot AI, one of China’s better-known artificial intelligence research labs, has set an ambitious commercial goal: Bloomberg reported the company is targeting about $2 billion in annualized revenue by the end of the year. That figure would represent roughly twice the revenue run rate Moonshot reported for August and follows strong early demand for its open-weight K3 model, launched this summer.
Despite a modest recent pullback in some usage metrics, third-party telemetry from OpenRouter indicates K3 models are still heavily used, with as many as 300 billion tokens generated each day on that system. Because Moonshot distributes K3s model weights openly, the company operates with slimmer margins than closed-weight providers, a structural difference that shapes both pricing and profitability.
Moonshots projected revenue remains small compared with the largest cloud AI providers: recent reports place OpenAI and Anthropic annual revenues near $40 billion and $65 billion, respectively. Still, Moonshots rising projections underscore that open-weight models can attract significant consumption and generate meaningful sales, even if they may not match the profits of proprietary frontier models.
The companys development practices have also drawn scrutiny. Anthropic has accused Moonshot of running a prolonged model distillation effort that allegedly routed nearly 300,000 requests from Moonshots Kimi system to Anthropics Claude Opus, and of collecting more than 23 million responses from Anthropic models for use in training. Those allegations have fueled controversy over how Moonshot builds and refines its models and whether its methods cross legal or ethical lines.
Keep Reading
TikTok misled consumers about filtering content for minors, judge rules
Nscale adds former OpenAI exec Fidji Simo to its board ahead of potential IPO
