Zhipu Debuts on HKEX, Valued at HK$52.8 Billion as First AGI Model Public Company

Zhipu, a Chinese artificial general intelligence (AGI) foundational model company, listed on the Hong Kong Stock Exchange (HKEX) on January 8, 2026, becoming the first publicly traded entity globally with AGI foundational models as its core business. The company's shares, trading under the ticker 02513.HK, opened at HK$120 per share, giving it a market capitalization of HK$52.828 billion.

Digital stock ticker showing Zhipu's HKEX code and initial trading price with an upward trend.
The stock surged during its debut, reaching a high of HK$129.9, an increase of approximately 11.8% from its offering price of HK$116.2. The Hong Kong public offering was oversubscribed 1159.46 times, while the international offering was oversubscribed 15.28 times. Zhipu's initial public offering (IPO) raised over HK$4.3 billion before the "greenshoe" option.
Cornerstone Investors and Strategic Vision
Zhipu attracted a lineup of cornerstone investors, including state-owned assets from Beijing, leading insurance funds, public funds, private equity funds, and industrial investors. Eleven cornerstone investors, such as JSC International Investment Fund SPC, JinYi Capital Multi-Strategy Fund SPC, and Perseverance Asset Management, collectively subscribed for HK$2.98 billion.

Portrait of Liu Debing, Chairman of Zhipu, in a professional setting.
Liu Debing, Chairman of Zhipu, stated during the listing ceremony that the company's long-term goal has been to "make machines think like humans." He highlighted the development of Zhipu's proprietary algorithmic architecture, GLM, launched in 2021. The recent release of GLM-4.7 has positioned the model as a global leader, according to Liu, and is foundational to the company's pursuit of AGI.
Technological Development and Market Position
Zhipu, spun off from Tsinghua University in 2019, has been an early developer of large models in China. The company pioneered the GLM general pre-training paradigm based on autoregressive infilling. It was the first in China to release a ten-billion-parameter model, an open-source hundred-billion-parameter model, a conversational model, a multimodal model, and the world's first device-controlling agent. The GLM architecture has achieved full localization, adapting to over 40 domestic chips.

Visual representation of Zhipu's GLM model evolution, showing growth from core to various advanced models.
According to a Frost & Sullivan report, Zhipu ranked first among independent general large model developers in China by 2024 revenue and second among all general large model developers. As of September 30, 2025, Zhipu's models powered 12,000 enterprise customers globally, with internet customers accounting for 50%. The models also reached over 80 million end-user devices and more than 45 million developers.
Financial Performance and R&D Investment
Zhipu's revenue grew from 57.4 million yuan in 2022 to 312.4 million yuan in 2024, representing a compound annual growth rate of 130%. In the first half of 2025, revenue reached 191 million yuan, a 325% increase year-on-year. Gross profit margins were 54.6% in 2022, 64.6% in 2023, 56.3% in 2024, and 50% in the first half of 2025.
The company's R&D investment totaled approximately 4.4 billion yuan between 2022 and the first half of 2025. R&D personnel constitute 74% of its workforce. This investment supports rapid technological iteration, with GLM series models undergoing foundational updates every 2-3 months.

Abstract depiction of interlocking gears and data streams, symbolizing continuous R&D and rapid technological iteration.
In December 2025, Zhipu's GLM-4.7 foundational model topped both open-source and domestic model categories in the Artificial Analysis comprehensive model capability ranking and the Code Arena coding ranking. Within two months of its release, 150,000 developers from 184 countries subscribed to its programming products, and over 50 domestic and international development platforms and tools, including Cerebras and Windsurf, integrated it.
Future Growth and Market Strategy
Approximately 70% of the offering's net proceeds, or HK$2.9 billion, will be allocated to R&D in general AI large models. Another 10%, or HK$420 million, will be used to optimize the company's Model as a Service (MaaS) platform, including providing foundational models, training/inference tools, and infrastructure.

Diagram illustrating Zhipu's Model as a Service (MaaS) platform connecting to various enterprise and developer services.
Zhipu began deploying its MaaS model in 2021, two years before the broader commercialization of large models. The platform offers model API calls, model subscriptions, and localized deployments. It currently serves over 3 million enterprises and application developers. Its programming subscription product has achieved over 100 million in annual recurring revenue (ARR) and has expanded into overseas developer communities.
Stay Ahead of the AI Curve
Join 50,000+ subscribers getting the latest AI tools, trends, and tutorials delivered to their inbox weekly.
No spam, unsubscribe at any time.