Zhipu ZHIPU: Why did it surge? Up 34% in a single day—China’s leading large-model stock rebounds from the bottom.

As of June 15, 2026, based on Gate market data, the Hong Kong stock AI sector collectively strengthened, with Zhipu (02513.HK) opening high in the morning and continuing to rise, with the highest intraday increase exceeding 47%, closing up 34%, becoming the top gainer in the Gate stock trading Hong Kong stock sector.

Since Gate officially launched Hong Kong stock trading services supporting over 1,000 Hong Kong stocks on June 11, 2026, this stock has become one of the platform's highly watched targets. This week, ZHIPU's sharp surge occurred after several days of consolidation, sparking market discussions on whether this "world's leading large model stock" has entered a trendful upward channel. The current rally is driven not by a single technical rebound but by multiple factors including industry geopolitical shifts, company product release pace, and business model validation.

How is Zhipu’s fundamental situation?

Zhipu Huazhang Technology Co., Ltd. was established in 2019, originating from the technological成果转化 of Tsinghua University’s Computer Science Department, and is one of the earliest teams in China to engage in the development of general artificial intelligence (AGI) large models. The company proposed the autoregressive fill-based general pre-training paradigm GLM, and has successively released China’s first 100-billion-parameter model, the first open-source 1-trillion-parameter model, the first dialogue model, and the first multimodal model. Zhipu’s core positioning is “building the brain”—focusing on自主研发 of general foundational large models, rather than shifting to downstream applications like world models or embodied intelligence.

In terms of commercialization path, Zhipu pioneered a standardized product system centered on MaaS (Model as a Service), delivering “general intelligent capabilities” to enterprises and developers via API calls, model subscriptions, and localized deployment, rather than following traditional enterprise AI routes with heavy delivery and strong customization. In 2024, Zhipu’s revenue ranked first among independent general large model developers in China, and second among all large model developers globally, with a market share of 6.6%.

Can Zhipu’s financial structure support its current market valuation?

According to Zhipu’s 2025 performance report, the company achieved an annual operating revenue of 724 million yuan, a year-on-year increase of 131.9%, making it the largest independent large model developer in China by revenue. From the revenue structure, local deployment revenue was 534 million yuan, up 102.3%, accounting for 73.7% of total revenue; cloud deployment revenue was 190 million yuan, up 292.6%, increasing its proportion from 15.5% in 2024 to 26.3%, indicating a shift in growth focus toward MaaS mode. Breaking down by business form, open platform and API revenue in 2025 was 190 million yuan, up 292.6%; enterprise-level intelligent agent revenue was 160 million yuan, up 248.8%; enterprise-level general large model revenue was 360 million yuan, up 70.5%.

However, high growth comes with high investment. In 2025, Zhipu recorded a net loss of 4.72B yuan, expanding 59.5% year-on-year; adjusted net loss was 3.18B yuan, with R&D expenses reaching 3.18B yuan, about four times the same period’s revenue. The growth in R&D costs mainly stems from two aspects: expanding the R&D team and increasing share-based payments; and paying third-party compute service providers. The overall gross margin decreased from 56.3% in 2024 to 41% in 2025, mainly due to the increased proportion of cloud deployment business and temporary decline in gross margin of local deployment.

As of the end of 2025, the company held approximately 2.26B yuan in cash and cash equivalents, with a debt-to-asset ratio of 15.2%, indicating a relatively light debt burden. The coexistence of high growth, high investment, and high losses is a core characteristic of Zhipu’s current financial structure.

What does the full release of GLM-5.2 mean for the large model product matrix?

On June 13, 2026, Zhipu announced the full release of its latest flagship model GLM-5.2, providing Lite/Pro/Max and team versions to all Coding Plan users, with an upcoming API interface. The model offers a context window of up to 1 million tokens, maintaining domestic leadership in long-range programming and agent tasks, with weights officially open-sourced under the MIT license. Early developer feedback indicates its coding ability is close to Claude Opus 4.8.

GLM-5.2 is the sixth-generation flagship model released intensively since 2025. Previously, the company launched GLM-4.5, GLM-4.6, GLM-4.7, GLM-5, and GLM-5.1, each setting new records in international evaluations. Notably, GLM-4.7 achieved an 84.9% SOTA score in the open-source domain on LiveCodeBench V6, outperforming Claude Sonnet 4.5. Zhipu’s technical spokesperson Zheng Qinkai explained that the core breakthrough of GLM-5 is that AI is no longer just a code-generating tool but has autonomous testing and iteration capabilities as an “engineer”; GLM-5.1 further realized a closed-loop capability for long-range task execution, capable of working continuously for over 8 hours in a single task.

From a commercial operation perspective, in 2025, Zhipu’s programming package “GLM Coding Plan” surpassed 242k paid developers, with token call volume increasing 15-fold in six months. Among China’s top ten internet companies, nine are deeply calling the GLM models daily. As of March 2026, Zhipu’s platform has over 4 million registered enterprises and users, serving over 218 countries and regions worldwide.

How does the export control event involving Anthropic reshape the domestic large model competition landscape?

On June 12, the U.S. Department of Commerce, under export control authorization, required Anthropic to suspend access to its latest flagship models Claude Fable 5 and Claude Mythos 5 for all foreign nationals. These two models were taken offline just three days after their release. Anthropic stated that the U.S. government’s decision was based on “serious misunderstandings,” but due to its inability to distinguish user nationality in real-time, it ultimately chose to suspend all users’ access.

This event has catalyzed a structural shift in China’s domestic large model industry. Orient Securities pointed out that the delisting of Anthropic models exposes the uncontrollable risks of closed-source models being limited by a single jurisdiction, favoring leading domestic model companies like Zhipu to form substitution effects. Huatai Securities believes that “domestic models + domestic compute power” have shifted from alternative options to essential choices, with enterprise demand for multi-model backups and localized deployment increasing exponentially.

Within less than 24 hours after the event, Zhipu quickly announced the full release of GLM-5.2, emphasizing that “technology should not belong only to a few, nor should it be revoked at any time.” The model is open-sourced under the MIT license, significantly lowering the threshold for enterprise localization deployment, which is expected to accelerate the penetration of domestic large models into enterprise markets.

How does Hong Kong stock AI sector benefit from import substitution logic?

On June 11, 2026, Gate officially launched Hong Kong stock trading services covering over 1,000 Hong Kong stocks, allowing users to trade Hong Kong stocks directly with USDT via a unified account, without currency conversion or additional KYC procedures. Zhipu, with today’s sharp rise, became one of the most prominent gainers in this sector. More importantly, the overall Hong Kong stock AI large model sector shows a resonant upward pattern.

From an index perspective, the Hang Seng Tech Index and Hong Kong Stock Connect completed component adjustments simultaneously on June 8, with Zhipu newly included in the Hang Seng Tech Index and also entering the Hong Kong Stock Connect, marking the official entry of AI large model companies into the core tech benchmark of Hong Kong stocks. Meanwhile, in 2025, China’s AI large model market size reached 242k yuan, a 49.1% YoY increase, with expectations to surpass 70 billion yuan in 2026. By the end of March 2026, China’s total intelligent compute power reached 1.88 million PFLOPS, providing infrastructure support for large-scale model applications. Regarding domestic large model API calls, data from OpenRouter, a global AI model API aggregation platform, shows that weekly calls for Chinese models have continued to surpass the U.S., accounting for over 33% of global total calls.

What does Zhipu’s accelerated A+H layout mean for the company’s strategy?

On June 1, 2026, Zhipu announced that it had applied for the issuance and listing of A-shares with relevant Chinese regulators and submitted an application to the Shanghai Stock Exchange for listing and trading on the STAR Market, just five months after its Hong Kong listing. According to the announcement, the A-share issuance will account for 2% to 8% of the total post-issuance share capital, aiming to raise 15 billion RMB, of which 12 billion RMB will be used for general foundation large model R&D, 2 billion RMB for building a one-stop MaaS platform, and 1 billion RMB for working capital.

The dual A+H capital platform holds multiple strategic implications. On the capital front, large model R&D is a high-investment, long-cycle hard technology track; in 2025, Zhipu’s R&D expenses exceeded 3.1 billion yuan. A-share fundraising will provide more ample capital support for subsequent compute expansion and model iteration. In terms of valuation, since listing on HKEX on January 8, 2026, with an issue price of HKD 116.2 per share, the stock closed at HKD 131.5 on the first day, with a market cap of about HKD 57.9 billion; as of today’s close, the share price has increased over 11 times from the listing, with a total market cap surpassing HKD 627.7 billion. In industry landscape, leading domestic large model companies like Zhipu, Kimi, DeepSeek are valued at record highs—DeepSeek at $150 billion, OpenAI at $850 billion, and Anthropic at $1.2 trillion. Listing in A-shares will help Zhipu further expand capital channels and enhance market influence, but the ultimate value realization still depends on technological commercialization and profitability.

How do institutions view Zhipu’s long-term investment value?

CMB International recently published a research report, initiating coverage on Zhipu with a “Buy” rating and a target price of HKD 1,503.9. The report notes that, based on 2024 revenue, Zhipu ranks first among China’s independent large model developers and second among all large model developers, with a market share of 6.6%, possessing full-stack capabilities from foundational model pretraining and fine-tuning to enterprise deployment and application development.

From a profitability perspective, Zhipu’s MaaS model is entering a scaled growth phase. The company’s API recurring revenue reached 1.7 billion yuan annually, up 60 times YoY, and gross margin of cloud deployment increased from 3.3% in 2024 to 18.9% in 2025. Miao Zhen, a partner at KPMG China, believes that the inflection point for large model profitability may arrive faster than in the internet era, with the industry possibly reaching a turning point within one to two years, making commercialization capability the key to victory.

What conditions need to be verified for trend reversal after a single-day surge?

The driving forces behind Zhipu’s sharp rise today are relatively clear: the full release of GLM-5.2 and open-source under MIT license provides product-level benefits; the Anthropic export control event opens up substitution market space; inclusion in Hong Kong Stock Connect and the Gate Hong Kong stock trading platform broadens investor coverage; and the strategic expectation of A+H listing further boosts market sentiment. These factors together form a multi-layered logical support for the current rally, distinguishing it from a pure technical rebound.

However, confirming a trend reversal still requires validation through several steps. First, whether the substitution demand triggered by the Anthropic event can be sustained and translated into actual API call volume and revenue growth for Zhipu needs to be verified in subsequent quarterly reports. Second, the industry-wide high loss pressure—Zhipu’s net loss in 2025 was about 6.5 times revenue—means that profitability clarity remains a core variable constraining long-term valuation. Third, the large model industry in 2026 is transitioning from “technology and scale competition” to “commercialization and monetization,” requiring ongoing monitoring of competitive landscape and market share changes.

Overall, Zhipu’s surge today is supported by multiple catalysts, but the sustainability of the trend needs more data on product penetration, enterprise client conversion, and gross margin improvement. For investors focusing on the Hong Kong stock AI sector, maintaining awareness of industry structural changes and technological iteration pace is crucial, as the short-term emotional-driven rise must be validated against long-term fundamentals over time.

Summary

Zhipu surged over 47 intraday today, closing up 34%, ranking first in the Gate stock trading Hong Kong sector. The core drivers include: the full release and open-source of the flagship model GLM-5.2 under MIT license; the substitution opportunities created by the Anthropic export control event; the inclusion in Hang Seng Tech Index and Hong Kong Stock Connect, improving liquidity; and the strategic outlook of A+H dual-capital layout. Fundamentally, Zhipu’s 2025 revenue was 724 million yuan, up 131.9%, making it the largest independent large model developer in China, but high R&D costs led to a net loss of 49.54B yuan. Industry-wise, China’s AI large model market is expected to surpass 70 billion yuan in 2026, with weekly call volume for domestic models already exceeding the U.S. The trend reversal confirmation requires subsequent validation of API call conversion, gross margin improvement, and profitability clarity.

FAQ

Q: What kind of company is Zhipu? Where is it listed?

Zhipu Huazhang Technology Co., Ltd. was founded in 2019, originating from Tsinghua University’s Computer Science Department’s technological成果转化, and is one of China’s earliest large model R&D firms, focusing on the development and commercialization of general AI foundational large models. Zhipu was listed on the Hong Kong Stock Exchange main board on January 8, 2026, known as the “world’s largest model stock,” with stock code 02513.HK.

Q: How was Zhipu’s financial performance in 2025?

In 2025, Zhipu achieved revenue of 724 million yuan, up 131.9% YoY, making it the largest independent large model developer in China by revenue. The net loss was 4.718 billion yuan, with R&D expenses around 4.72B yuan. The overall gross margin was 41%. The company held about 4.72B yuan in cash and equivalents, with a debt ratio of 15.2%, indicating a relatively light debt load.

Q: What improvements does GLM-5.2 have over previous versions?

GLM-5.2 is Zhipu’s most capable open-source large model to date, supporting a truly usable 1 million token context window, maintaining domestic leadership in long-range programming and agent tasks. Early developer feedback shows its coding ability is close to Claude Opus 4.8. The model is open-sourced under the MIT license, greatly lowering the threshold for enterprise localization deployment.

Q: What impact does the Anthropic export control event have on Zhipu?

The U.S. requiring Anthropic to suspend access for non-U.S. users to its flagship models exposes the uncontrollable risks of closed-source models being limited by a single jurisdiction. This accelerates enterprise demand for domestic models as substitutes. Huatai Securities pointed out that “domestic models + domestic compute power” have shifted from optional to essential, with enterprise demand for multi-model backups and localized deployment increasing exponentially, benefiting companies like Zhipu.

Q: What does Zhipu’s A+H layout imply?

On June 1, 2026, Zhipu announced applying for A-shares issuance and listing on the STAR Market, aiming to raise 15 billion RMB, with 12 billion RMB allocated to general foundation large model R&D. This dual-capital approach helps expand capital channels, supports further R&D, and enhances market influence, aligning with long-term strategic development.