The Hong Kong Stock Exchange (HKEX) is witnessing the dawn of a new AI era.

Since the beginning of 2026, the capital market for China’s Large Language Model (LLM) industry has been exceptionally vibrant. As the "world’s first LLM stock," Knowledge Atlas Tech (02513.HK) staged a listing debut where its market capitalization skyrocketed from HK$50 billion to over HK$320 billion in just 43 days.

The market value ascended like a rocket, even briefly surpassing some traditional internet giants that have been deeply rooted in the market for twenty years. On February 20, 2026, Knowledge Atlas Tech's share price rose from its offer price of HK$116.2 to HK$725, representing a cumulative increase of 524% and reaching its highest point since listing.

**An Apology Letter Amidst Breaking the HK$300 Billion Market Cap Threshold**

NewTimeSpace has learned that Knowledge Atlas Tech only debuted on the HKEX on January 8, 2026. At that time, the offer price was only HK$116.2, but in just over a month, it surged to a peak of HK$725, a fivefold increase. It has become the most brilliant performer in the domestic LLM track since the start of the Year of the Horse.

However, as Knowledge Atlas Tech’s share price set a historic record on February 20, 2026, an apology letter followed closely behind.

On the evening of February 21, 2026, Knowledge Atlas Tech issued an apology letter to its developer community, publicly apologizing to users for the insufficient capacity expansion of the GLM Coding Plan and announcing a full refund and additional compensation scheme.

It is understood that on February 12, 2026, Knowledge Atlas Tech officially released its next-generation flagship model, GLM-5. According to officially disclosed information, this model has reached a leading level in the open-source field regarding Coding and Agent capabilities, with performance in complex system engineering and long-range task processing approaching that of the world-renowned model Claude Opus 4.5.

The appearance of this apology letter came only nine days after the official release of GLM-5.

Furthermore, on the day following the release of GLM-5, Knowledge Atlas Tech became the first AI enterprise in China to raise prices for its commercial LLM services. Its package prices in the China region rose by 30%, while the overseas version saw an increase of over 100%.

While Knowledge Atlas Tech’s technological upgrades deserve a reasonable return, the stability of the company’s services failed to improve in tandem with the price hike. This resulted in many users paying higher fees only to encounter restricted services, which triggered a PR crisis for Knowledge Atlas Tech.

Following the company's crisis, on the evening of February 24, 2026, Haizhi Technology Group announced on the HKEX that it had recently signed a new round of strategic cooperation framework agreements with Knowledge Atlas Tech. The two parties will focus on strategic cooperation in fields such as model training and industrial scenario implementation, relying on Haizhi Technology's graph-model fusion technology (such as the Atlas graph) and Knowledge Atlas Tech’s LLM capabilities to achieve deep complementarity.

Under the influence of multiple positive factors, Knowledge Atlas Tech's share price is currently in a state of high-level fluctuation. As of the close on February 26, 2026, the company was quoted at HK$556.5, with a total market capitalization of HK$248.1 billion.

However, whether Knowledge Atlas Tech can sustain a market value of over HK$200 billion will depend on the company's future data and performance.

**Star Institutions Provide Support, but Self-Sustaining Capabilities Remain Insufficient**

Public information shows that Knowledge Atlas Tech was founded in 2019 under the full name Beijing Zhipu Huazhang Technology Co., Ltd. It originated from the transformation of technological achievements from the Knowledge Engineering Laboratory of the Department of Computer Science at Tsinghua University and is among the first batch of independent domestic manufacturers dedicated to general-purpose LLMs.

In 2022, Knowledge Atlas Tech released GLM-130B, the first domestic open-source LLM with 100 billion parameters. The launch of ChatGLM-6B in 2023 allowed ordinary developers to run large models on home computers, directly lowering the industry threshold.

Currently, Knowledge Atlas Tech has formed a complete product matrix, including AI assistants for ordinary users, a MaaS (Model-as-a-Service) platform for developers, and private solutions for government and enterprise customers, moving forward on the "two legs" of technology and commercialization.

Moreover, Knowledge Atlas Tech has gathered more than 50 investment institutions. Before listing, its cumulative financing exceeded RMB 8.3 billion, involving top-tier firms like Legend Capital, Qiming Venture Partners, and Sequoia China, as well as strategic investments from giants like Meituan, Ant Group, and Tencent, alongside various local state-owned capital and overseas funds.

In the cornerstone investment phase at the end of December 2025, Knowledge Atlas Tech attracted as many as 11 investors, with a total financing amount of HK$2.984 billion. Cornerstone investors accounted for a high subscription proportion of 68.6%.

However, a look at Knowledge Atlas Tech’s prospectus reveals a set of "suffocating" figures: the company's self-sustaining capability is severely insufficient. Data shows that from 2022 to the first half of 2025, the company’s cumulative R&D investment exceeded RMB 4.4 billion, while cumulative revenue for the same period was only RMB 685 million.

In the first half of 2025, R&D investment was RMB 1.595 billion, which is 8.35 times the revenue of RMB 191 million. During the same period, the loss was RMB 2.358 billion, and the ratio of loss to revenue soared from 252.3% in 2022 to 1,234.6% in the first half of 2025. The adjusted net loss was RMB 1.752 billion, with the magnitude of the loss widening year-on-year. This means that for every RMB 1 of revenue earned, the company burns RMB 8.

It is worth mentioning that in 2024, Knowledge Atlas Tech's R&D investment reached as high as RMB 2.1954 billion, of which the cost of purchasing computing power accounted for 70.7%, amounting to RMB 1.553 billion. In the first half of 2025, this proportion remained above 70%.

Currently, how to balance technological investment with commercial returns is the Sword of Damocles hanging over Knowledge Atlas Tech.

**Intense LLM Competition; Knowledge Atlas Tech’s Operational Capabilities Need Strengthening**

The current domestic LLM market is undergoing a brutal shakeout from the "War of a Hundred Models" to the "Top Five Base Models." To date, models such as Alibaba (Tongyi Qianwen), ByteDance (Doubao), DeepSeek, and Yuanbao have taken the lead by virtue of their computing power reserves and data advantages. Other models like Knowledge Atlas Tech and MiniMax are left to compete for the remaining market share.

The squeeze from giants is reflected not only at the technical level; Alibaba and Tencent control traffic distribution channels, causing the largest profit pools to flow to platforms rather than model companies.

Presently, the performance of models like ByteDance’s Doubao 2.0 and Alibaba’s Qwen 3.5 is approaching that of Knowledge Atlas Tech, but by utilizing their own computing clusters, their inference costs are only 60% of Knowledge Atlas Tech's. In the price war, Knowledge Atlas Tech's GLM-4.5 is priced at RMB 0.5 per million tokens, while DeepSeek-V3 has dropped to RMB 0.1, representing a vast disparity.

Starting in 2024, Knowledge Atlas Tech accelerated its transformation toward a MaaS (Model-as-a-Service) platform, attempting to replicate the success of the American AI unicorn Anthropic by achieving "asset-light, high turnover, and wide coverage" through standardized API calls.

The stronger the model, the scarcer the tokens; thus, price increases align with industry logic. However, Knowledge Atlas Tech’s commercialization team was clearly unprepared to handle the traffic. After the release of GLM-5, because the traffic far exceeded expectations and the company’s computing power expansion could not keep up, the company was forced to apologize and offer compensation.

Behind this incident, what appeared to be a technical "outage" also indirectly exposed the "shortcomings" in the company's operational capabilities.

For Knowledge Atlas Tech, listing is not the finish line but the starting point of a more rigorous test. Knowledge Atlas Tech needs to prove to the capital market that it can achieve "self-sustenance" in the future and support a market capitalization of hundreds of billions.

NewTimeSpace Disclaimer: All content herein is the original work of NewTimeSpace. Any reproduction, reprinting, or use of this content in any other manner must clearly indicate the source as "NewTimeSpace". NewTimeSpace and its authorized third-party information providers strive to ensure the accuracy and reliability of the data, but do not guarantee the absolute correctness thereof. This content is for reference only and does not constitute any investment advice. All transaction risks shall be borne by the user.