Author｜Yang Licheng

Editor｜Wang Yimo

On June 17, the China Securities Regulatory Commission (CSRC) announced the completion of the regulatory review for Beijing Zhipu Huazhang Technology Co., Ltd.'s (Zhipu) Initial Public Offering (IPO) on the STAR Market. This marks a significant stride forward for large AI models in re-entering China's A-share market.

Regulatory Review Finalized

The CSRC Beijing Regulatory Bureau updated its public辅导 (tutoring, hereinafter referred to as "sponsor") disclosure platform, officially changing Zhipu's STAR Market IPO sponsor status to 'review completed.' Guotai Junan Securities serves as the lead sponsor for this listing.

This milestone indicates that Zhipu has concluded essential pre-listing preparations, including equity restructuring, compliance management, and financial standardization. It brings the company a step closer to formally submitting its prospectus to the Shanghai Stock Exchange.

Tracing its capitalization journey, Zhipu has consistently pursued a clear A+H dual-listing strategy. In April 2025, the company initiated A-share listing sponsorship with the Beijing Regulatory Bureau, later prioritizing a Hong Kong IPO. On January 8, 2026, Zhipu made its debut on the Hong Kong Stock Exchange's main board (Stock Code: 02513.HK), becoming the world's first independently listed general-purpose large model enterprise, with an initial market capitalization exceeding HK$51 billion.

Within six months of its Hong Kong listing, Zhipu promptly resumed its A-share listing process. In February 2026, it withdrew its original sponsorship filing and restructured its sponsorship framework, engaging Guotai Junan Securities and CICC as joint sponsors.

On June 1, the board officially approved the STAR Market listing plan, proposing to issue shares representing 2%-8% of the total equity post-issuance, aiming to raise approximately 15 billion yuan in total funds.

According to disclosures, the proceeds will be predominantly allocated to core operations: 12 billion yuan for research and development (R&D) of general-purpose AI foundation models, including technological advancements in next-generation GLM series models and computing cluster construction; 2 billion yuan for upgrading the Model-as-a-Service (MaaS) one-stop platform to enhance developer ecosystems and commercialization frameworks; and the remaining 1 billion yuan to bolster working capital, ensuring financial security for R&D and operations.

As a rare pure-play general large model entity in China, Zhipu's return to the A-share market has garnered significant market attention. Its STAR Market debut will fill the void for authentic large model core assets in China's A-share market.

Revenue Growth Amidst Expanding Net Losses

Founded in 2019 through technology transfer from Tsinghua University's Knowledge Engineering Group (KEG), Zhipu is led by Professor Tang Jie's core R&D team, one of China's earliest pioneers in general large model commercialization.

According to Zhipu's Hong Kong prospectus and 2025 annual report, its core assets encompass the fully self-developed GLM series general foundation models, covering language, code, multimodal generation, and intelligent agent technologies.

Zhipu operates a consumer-facing product, 'Zhipu Qingyan,' and a B2B MaaS open platform, with commercialization divided into localized private deployment and cloud-based API services.

Financially, Zhipu has sustained triple-digit revenue growth: 57.409 million yuan in 2022, 125 million yuan in 2023, and 312 million yuan in 2024 (Compound Annual Growth Rate [CAGR]: 130%). In 2025, revenue surged to 724 million yuan, up 131.9% year-over-year (YoY).

By 2024 revenue, Zhipu ranked first among China's independent general large model vendors, holding a 6.6% market share.

Its business structure features 'private deployment as the anchor, cloud services as the growth engine.'

In 2025, localized private deployment generated 534 million yuan (73.7% of total revenue), primarily serving state-owned enterprises, financial institutions, and energy firms with stringent data security requirements. Project renewal rates reached 95%, forming the core revenue and profit pillar.

Cloud-based MaaS revenue hit 190 million yuan, soaring 292.6% YoY—significantly outpacing overall growth.

By September 2025, Zhipu had served over 12,000 global enterprise clients, empowered over 80 million terminal devices, and attracted over 45 million developers, establishing a preliminary developer ecosystem.

As of the first quarter of 2026, Zhipu's MaaS platform annual recurring revenue (ARR) reached 1.7 billion yuan, with sustained momentum in cloud business growth.

Rapid growth coincides with industry-wide high-input losses. Prospectus and annual report data reveal net losses of 144 million yuan in 2022, 788 million yuan in 2023, 2.958 billion yuan in 2024, and 4.718 billion yuan in 2025.

Losses primarily stem from intensive R&D and computing investments. In 2024, R&D spending reached 2.195 billion yuan, rising to 3.180 billion yuan in 2025, with computing procurement costs forming the core of R&D expenditure.

Gross margin stood at 41.0% in 2025, with private deployment margins significantly higher than cloud services—remaining within reasonable ranges for the large model industry.

Revenue Falls Short of Covering Costs

Industry trends indicate a pivotal shift from 'parameter competition' to 'commercialization,' with sustained growth dividends.

Frost & Sullivan data shows China's large language model market reached 5.3 billion yuan in 2024, projected to grow to 101.1 billion yuan by 2030 (CAGR: 63.7%).

Enterprise demand will dominate nearly 90% of the market, driven by government/enterprise digital transformation, vertical industry model customization, and AI agent applications.

The industry has evolved from an early 'hundred-model competition' to a consolidated phase featuring differentiated competition between internet giants and independent large model vendors.

Baidu, Alibaba, ByteDance, and other tech conglomerates leverage cloud infrastructure and consumer-end (C-end) traffic to dominate the general cloud market.

Independent vendors like Zhipu, by virtue of technological neutrality and full-stack self-development capabilities, deepen government/enterprise private deployment sectors, building differentiated competitive barriers.

Zhipu's core value lies in its fully autonomous technological foundation. The GLM series achieves full-stack self-research in pre-training frameworks and core algorithms, fully compatible with domestic computing power and information technology innovation systems. This precisely meets data security and self-sufficiency requirements for government and state-owned clients—the rationale behind its surging government/enterprise orders.

A STAR Market listing would provide Zhipu access to local capital for computing procurement and talent expenses required for model iteration, ensuring sustained R&D of next-gen foundation models.

Listing on the A-share market would enhance industry visibility, deepen government/enterprise client trust, and expand enterprise business reach.

Leveraging hard-tech valuation premiums to adjust the shareholder structure would hedge against uncertainties from single-market fluctuations. Amid import substitution trends, the long-term strategic value of China's leading large model enterprises will continue to rise.

However, industry expansion conceals persistent risks. Prolonged heavy R&D spending leaves revenue unable to cover costs. Even with this fundraising buffering short-term gaps, the unproven profit model necessitates continuous external financing.

Industry intensification sees internet platforms slash service pricing through computing scale advantages, squeezing online business margins.

Established tech firms competing in vertical sectors vie for government/enterprise orders, further intensifying market competition.

Additional variables include geopolitical risks in high-end computing supply chains, uncertainties in AI technological evolution, stricter data compliance/content regulation, and fluctuations in top client revenue contributions.

In the short term, Zhipu can sustain revenue growth through private deployment orders. Long-term profitability hinges on increasing standardized cloud product adoption, reducing unit computing costs, and achieving scalable commercial closed loops.

Zhipu's completion of the STAR Market regulatory review marks a new phase in China's large model capitalization.

The A-share market will soon welcome its first pure-play general large model entity. This listing will determine the company's future R&D iteration and market expansion pace while providing a reference model for domestic AI hard-tech enterprises.

Capital market valuation will continue to focus on technological strength, monetization capabilities, and long-term competitive barriers.