China's Largest Single Funding Round for Large Model Startups: Moonshot AI Secures Over $1 Billion in Financing
-
AI startup Moonshot AI recently completed a funding round exceeding $1 billion, with investors including Sequoia China, Xiaohongshu, Meituan, and Alibaba, while existing shareholders also participated in follow-on investments. This funding round has elevated Moonshot AI's valuation to approximately $2.5 billion, positioning it as one of the leading companies in the domestic large model sector. This marks the largest single funding round secured by a Chinese AI large model company since ChatGPT sparked a global frenzy.
Moonshot AI's previous funding round in 2023 raised over $200 million, with investors including Sequoia China and ZhenFund.
Founded in March 2023, Moonshot AI quickly emerged as a significant player in the large model field. Its core team members have contributed to the development of various large model projects such as Google Gemini, Google Bard, PanGu NLP, and WuDao, bringing years of research and development experience in large models. Currently, the company has a team of over 80 members.
According to official introductions, Moonshot AI is a next-generation cross-modal large model developer, dedicated to researching and developing advanced cross-modal large models. The company has pioneered RoPE (Rotary Position Embedding), a critical component in mainstream models like Meta's LLaMa and Google's PALM, and introduced group normalization, a key element in the success of AI models such as Stable Diffusion. Yang Zhilin, the company's founder, is a post-90s academic prodigy who graduated from Tsinghua University's Computer Science Department under the guidance of Professor Tang Jie. He later earned his Ph.D. in Computer Science from Carnegie Mellon University, studying under Ruslan Salakhutdinov, Apple's current AI head and one of the pioneers of deep learning. Yang has conducted research in natural language processing at Facebook AI Research and Google Brain, winning first place in the DREAM9 Global Cancer Prediction Competition and second place globally in Alibaba's Tianchi Big Data Competition, as well as the 2017 Nvidia Pioneer Research Award. He has published over twenty papers at top-tier AI conferences such as ICLR, NIPS, ICML, KDD, and ACL, and holds world-leading positions (State-of-the-art) in all six mainstream language modeling datasets.
In June 2023, tech media The Information listed Yang Zhilin as one of the five leading candidates for "China's OpenAI," a testament to his high recognition in the industry. The other four candidates are MiniMax, Zhipu AI, Light Years Beyond, and Langchao Tech.
Since its inception, Yang's company, Moon Dark Side, has achieved a comprehensive layout from general large-scale models to upper-layer applications in less than a year. The company has trained a self-developed general large-scale model with hundreds of billions of parameters and launched the consumer-facing Kimi AI Assistant in October 2023, marking its first foray into a To C super app.
Kimi AI Assistant supports long-text input of up to 200,000 Chinese characters, emphasizing lossless memory. "Long Context" is one of Moon Dark Side's flagship technologies, stemming from the team's ambition to overcome the implementation bottlenecks of large models. The "emergence" of large models' intelligence is primarily due to scaling up parameters to the hundreds of billions. In November, the company announced that its Kimi Chat chatbot service is now open to the public, accessible through the official website: https://kimi.moonshot.cn/.
Yang Zhilin believes that in the B2C sector, AI Native will create new traffic channels and has the potential to become the Super App of the AI era, offering significant business opportunities.
It is reported that Moonshot AI is currently developing a general multimodal model, expected to launch within this year. Multimodal models have become a key competitive focus among major model startups recently.