Beijing-based Moonshot AI has closed approximately $2 billion in new funding at a $20 billion valuation, signaling sustained investor appetite for open-weight large language models even as inference cost advantages drive adoption. The round was led by Meituan’s VC arm, Long-Z Investment, with participation from Tsinghua Capital, China Mobile, and CPE Yuanfeng.

This brings Moonshot’s total capital raised over the past six months to $3.9 billion. The company’s valuation trajectory has been steep: from $4.3 billion at the end of 2025 to $10 billion after a $700 million raise in early 2026, and now $20 billion. Annual recurring revenue surpassed $200 million in April, fueled by paid subscriptions and API usage.

Moonshot’s Kimi K2.5 model, released earlier this year, achieved near-benchmark performance against OpenAI and Anthropic models, making it a serious contender in code generation. The latest iteration, Kimi K2.6, is the second most-used LLM on OpenRouter, a distribution platform that routes inference requests across providers.

For data and ML engineers, the key takeaway is the trade-off between performance and cost. Open-weight models from Chinese labs offer competitive inference economics—often at a fraction of the per-token cost of proprietary Western models—while closing the capability gap. This makes them attractive for high-throughput production workloads where latency budgets are tight but cost sensitivity is higher.

Moonshot faces stiff competition from DeepSeek (reportedly raising at $45B), publicly traded Zhipu AI ($55.9B market cap), and MiniMax ($33B). Backers include Alibaba, Tencent, HongShan, ZhenFund, IDG Capital, and 5Y Capital.

Source: TechCrunch