SHANGHAI — Knowledge Atlas Technology, better known as Zhipu or Z.ai, made a solid debut on the Hong Kong Stock Exchange this week, marking a milestone for China’s generative artificial intelligence industry as the first major large language model (LLM) developer to go public.
Shares of the Beijing-based company climbed as much as 16 per cent above their offer price of HK$116.20, before paring gains to close the session up about 13 per cent. The IPO raised roughly US$558 million, valuing the company at around US$6.6 billion and drawing heavy interest from retail and institutional investors alike.
Zhipu’s listing is being closely watched as a bellwether for China’s so-called “AI tigers” — start-ups racing to build foundational models capable of competing with US leaders such as OpenAI and Anthropic. Its public market debut offers an early signal of how investors view the long-term commercial prospects of Chinese generative AI firms, which operate under very different constraints from their Silicon Valley peers.
Founded in 2019 by researchers from Tsinghua University, Zhipu has positioned itself as a pioneer in China’s large-model ecosystem. The company develops the GLM family of LLMs, including open-weight models that allow enterprise users to customise systems for specific tasks. Backed by major Chinese technology groups including Alibaba and Tencent, as well as state-linked funds, Zhipu has secured contracts with state-owned enterprises that favour tailored, on-premise AI infrastructure over public cloud solutions.
Despite the strong debut, the broader challenges facing Chinese AI software companies remain evident. US export controls continue to restrict access to advanced semiconductors and related expertise, limiting training capacity and driving up costs. Compared with US rivals, Chinese developers also operate with significantly less capital and computing power, while competing in a fiercely crowded domestic market that pressures pricing.
“The Chinese market is hyper-competitive, which naturally drags prices down,” Zhipu co-founder and chairman Liu Debing said in a recent interview. “But as we compete globally, international users will recognise the value.”
Financially, Zhipu is still in an investment-heavy phase. The company reported revenue of 312.4 million yuan in 2024, but posted sizable losses driven by aggressive research and development spending. Around 70 per cent of the IPO proceeds are earmarked for further work on general-purpose large models, rather than near-term commercial expansion.
Zhipu is listed under Hong Kong’s Chapter 18C regime for specialist technology companies, a framework designed to attract high-growth but unprofitable firms, and one that comes with higher volatility and valuation uncertainty. Analysts note that this differentiates AI software listings from recent Chinese semiconductor IPOs, which have benefited from clearer demand signals tied to domestic chip localisation.
The debut comes amid a flurry of AI-related listings in Greater China. Several chip designers and GPU makers have surged on their first trading days, while rival LLM developer MiniMax is preparing its own Hong Kong flotation. Together, the deals underscore Beijing’s push to strengthen domestic capabilities across the AI supply chain, from hardware to foundational software.
For now, Zhipu’s successful IPO provides a boost to sentiment around Chinese generative AI — but it also highlights the stark valuation gap with US counterparts, some of which are seeking funding rounds at valuations many times larger. Whether public markets will continue to reward China’s AI tigers will depend on their ability to scale globally, navigate geopolitical headwinds, and translate technical progress into sustainable revenues.
10-01-2026
