The race for China's first large language model (LLM) IPO has begun. Zhipu AI, a Beijing-based artificial intelligence company, has passed the Hong Kong Stock Exchange's listing hearing and officially submitted its prospectus. Almost simultaneously, another major LLM player, MiniMax, also cleared the hearing—marking the start of a fierce competition for the title of "China's first LLM stock."
Founded in 2019, Zhipu AI boasts a strong Tsinghua University pedigree. CEO Zhang Peng holds bachelor's, master's, and doctoral degrees from Tsinghua, while Chairman Liu Debing is also a Tsinghua alumnus. The company has attracted an impressive lineup of investors.
This development follows another heated race earlier this month—China's first domestic GPU IPO—with Moore Threads and MetaX preparing to go public, promising substantial returns for venture capital. Now, a similar scenario is unfolding in China's LLM sector, where an IPO represents a crucial milestone for investors to realize gains from the country's AI boom.
Zhipu AI traces its roots to Tsinghua's Knowledge Engineering Group (KEG) Lab, where its foundational technology was developed. Zhang Peng joined the lab in 2002 as a master's student after completing his undergraduate studies at Tsinghua. In 2019, Zhang and his team decided to commercialize their research, founding Zhipu AI.
The company hit the ground running. Zhang noted that Zhipu AI launched with established technology, a ready team, and existing clients, generating revenue from day one. According to its prospectus, Zhipu AI positions itself as a leader in China's AI sector, focusing on artificial general intelligence (AGI). In 2021, it released China's first proprietary pre-trained LLM framework, GLM, and introduced its Model-as-a-Service (MaaS) platform.
Today, Zhipu AI offers a full-stack model matrix covering language, code, multimodal, and agent capabilities, compatible with over 40 domestic chips. By June 2025, it had open-sourced more than 50 models, with total downloads exceeding 45 million.
Revenue comes primarily from B2B services via its MaaS platform, including localized and cloud-based deployments. Zhipu AI reported revenues of RMB 57.4 million (2022), RMB 124.5 million (2023), and RMB 312.4 million (2024), achieving a 130% CAGR. H1 2025 revenue reached RMB 190.9 million.
However, heavy R&D investments have led to mounting losses: RMB 144 million (2022), RMB 788 million (2023), RMB 2.958 billion (2024), and RMB 2.358 billion (H1 2025). Cumulative R&D spending stands at approximately RMB 4.4 billion.
Zhipu AI's funding journey reflects China's LLM investment frenzy. Early backers included CAS Star and others, followed by a RMB 152 million Series A. Subsequent rounds attracted major players like Meituan, Ant Group, Alibaba, Tencent, Xiaomi, Sequoia China, and Hillhouse Capital. Its latest Series B6 round raised RMB 4.377 billion, with participation from state-backed funds. To date, Zhipu AI has completed eight funding rounds totaling over RMB 8.3 billion, reaching a valuation of RMB 24.377 billion.
The urgency for IPO is clear. MiniMax, another LLM contender, also filed for a Hong Kong listing. Founded in 2021 by Yan Junjie, a Tsinghua postdoc and former SenseTime researcher, MiniMax boasts backers including Hillhouse, miHoYo, and Tencent, with a latest valuation of $4 billion.
Hong Kong's more accommodating listing requirements for pre-profit tech firms make it an attractive destination compared to mainland China's stricter A-share market. The recent success of DeepSeek has further boosted interest in Chinese AI listings.
As China's LLM sector enters its "mid-game," commercialization and capital markets are taking center stage. Industry observers predict a coming polarization, where listed leaders will consolidate advantages while smaller players face acquisition or exit.
The Wall Street Journal recently reported that Moonshot AI aims for a 2026 IPO after closing a new funding round at a $4 billion valuation. In this high-stakes race, no contender can afford to fall behind.
Comments