Moonshot AI targets €1.84B revenue as open-weight K3 model gains traction
Beijing-based Moonshot AI aims to double annualized revenue to €1.84 billion ($2 billion) by end of 2026, driven by its open-weight Kimi K3 model. The company hit €920 million ($1 billion) in August, up from €276 million ($300 million) in June, per Bloomberg. The target tests whether open-weight AI can generate large-scale commercial revenue.
Bottom line — Moonshot's €1.84B target challenges the economics of closed-weight AI labs like OpenAI and Anthropic.
Go deeper (9)
- Moonshot's K3 model, released in July, has a 2.8 trillion parameter mixture-of-experts architecture and a 1 million token context window, according to the company's technical materials.
- OpenRouter data cited by TechCrunch shows K3 models generating up to 300 billion tokens daily, though usage has declined slightly in recent months.
- Moonshot's API pricing undercuts US rivals: €2.76 ($3) per million input tokens and €13.80 ($15) per million output tokens, with cached input at €0.28 ($0.30) per million tokens.
- Anthropic accused Moonshot of model distillation, alleging that nearly 300,000 requests from Kimi were routed to Claude Opus and over 23 million responses collected for training, per TechCrunch.
- Moonshot confidentially filed for a Hong Kong IPO in September, targeting a valuation above €46 billion ($50 billion), according to Bloomberg and Caproasia.
- The company raised €1.84 billion ($2 billion) in May 2026 at a €18.4 billion ($20 billion) valuation, led by Meituan's VC arm, per TechCrunch.
- Moonshot reportedly used Nvidia Blackwell chips to train K3 despite US export restrictions, accessing compute through Chinese firms with equipped data centers, per The Information via Tom's Hardware.
- Moonshot's revenue trajectory: €184 million ($200 million) ARR in April, €276 million ($300 million) in June, then €920 million ($1 billion) in August after K3's launch.
- The company's projected revenue is dwarfed by OpenAI (€36.8 billion/$40 billion) and Anthropic (€59.8 billion/$65 billion), but its open-weight model has lower margins, per TechCrunch.