Daily Report on Hong Kong-Listed LLM Concept Stocks (June 23, 2026)

On June 23, Hong Kong stocks suffered a broad decline, triggering a sharp correction in the LLM and semiconductor sectors. MINIMAX plunged over 16% ahead of its impending lock-up expiration, while ZHIPU AI dropped 10%. On the industry front, both infrastructure development and commercial applications accelerated: Intel launched its "AI Home Brain" with domestic tech giants to propel the smart home ecosystem; Volcano Engine revealed Doubao's daily volume exceeded 180 trillion tokens alongside the release of Doubao 2.1 Pro; and DeepSeek established a dedicated "Harness" group to systematically build out its Agent infrastructure and product capabilities.

I. Industry Observations

On June 23, 2026, Hong Kong stocks suffered a broad decline. The Hang Seng Index (HSI) closed down 1.82% at 23,336.28 points, the Hang Seng TECH Index fell 3.30% to 4,399.22 points, and the Hang Seng China Enterprises Index (HSCEI) dropped 1.96% to 7,759.36 points. Total daily turnover stood at HKD 334.361 billion, with Southbound trading recording a net inflow of HKD 10.371 billion.

The AI and LLM supply chains underwent a significant correction. MINIMAX-W plunged 16.46% as the company approaches its initial lock-up expiration on July 9. KNOWLEDGE ATLAS dropped 9.96%, and Bilibili-W shed 6.1%. 51WORLD declined 15.1%, and Unisound fell 14.4%.

The semiconductor and computing power sectors exhibited divergence. Nations Technologies advanced 10.9% amid widespread price hikes in the MCU industry and the company's continued tracking of cutting-edge Co-Packaged Optics (CPO) technology. Zhiyuan Holdings rose 11.9% driven by its forward-looking layout in the computing power domain. On the downside, Montage Technology tumbled 12.4%, Victory Giant Technology dropped 12.2%, while Hua Hong Semiconductor edged up 1.0%.

II. Market Performance of LLM Concept Stocks

Stock Code

Stock Name

Latest Price (HKD)

Change

03690.HK

MEITUAN-W

69.6

-3.33%

09888.HK

BIDU-SW

104.7

-3.68%

09988.HK

BABA-W

98.95

-3.84%

00020.HK

SENSETIME-W

1.39

-4.14%

00700.HK

TENCENT

414.8

-4.20%

03317.HK

XUNCE

124.2

-4.83%

01024.HK

KUAISHOU-W

43.22

-5.05%

02718.HK

MININGLAMP-W

223

-7.62%

02513.HK

KNOWLEDGE ATLAS

2,170.00

-9.96%

00100.HK

MINIMAX-W

515

-16.46%

Data Source: HKEX. For reference only, not investment advice. Data as of market close onJune 23, 2026. 

III. Industry and Corporate Dynamics

1. Intel Launches "AI Home Brain" Solution, Partnering with Domestic Giants Including MiniMax, Xiaomi, and Lenovo

On June 22, Intel unveiled its "AI Home Brain" solution at the launch event of the 4th China International Supply Chain Expo (CISCE), positioning it as a cornerstone for a new, agent-driven smart home ecosystem. Domestic technology pioneers including Haier, Hisense, Lenovo, Tencent, Midea, TCL, JD.com, ModelBest, Xiaomi, Honor, Tuya Smart, and MiniMax attended the launch conference. The solution is engineered to drive the commercial deployment of advanced AI technologies into smart home environments via extensive ecosystem collaborations.

2. Volcano Engine: Doubao LLM's Daily Volume Exceeds 180 Trillion Tokens; Unveils Doubao 2.1 Pro

On June 23, at the 2026 FORCE Conference, ByteDance's cloud division, Volcano Engine, disclosed that the average daily token usage of its Doubao LLM has surpassed 180 trillion tokens. This represents a staggering 1,500-fold surge over the past two years compared to its initial baseline of 120 billion tokens in May 2024. Volcano Engine currently commands a dominant 49.5% token market share in China's public cloud MaaS (Model-as-a-Service) sector. Concurrently, the company rolled out Doubao 2.1 Pro, which delivers monumental capability leaps across three core domains—coding, agents, and Vision-Language Models (VLMs)—while slashing comprehensive utilization costs by nearly 80% compared to Claude Opus 4.6. Volcano Engine also previewed its proprietary video generation model, Seedance 2.5, which is scheduled for an official release in July. Currently, the number of enterprise clients executing an annual token call volume exceeding 1 trillion has reached 200.

3. DeepSeek Establishes "Harness" Group to Scale Up Recruitment for "Model + Harness = Agent" Ecosystem

On June 23, DeepSeek's newly established "Harness" group accelerated its talent acquisition pipeline, with team lead Cui Tianyi publicly posting openings for Harness researchers, engineers, and product managers. Operating under the strategic paradigm of "Model + Harness = Agent," the group's research and development focuses on context management, long-term memory, sub-agents and multi-agent frameworks, self-evolving agents, real-world task feedback loops, and user feedback optimization for Agent products. This initiative marks a major milestone as DeepSeek systematically deploys capital and engineering resources toward Agent infrastructure and productization capabilities, moving well beyond the layer of foundational models alone.

IV. Introduction to LLM Concept Stocks

Stock Name

Stock Code

Market Position

LLM Products & Business Profile

KNOWLEDGE ATLAS

02513.HK

A leading player in China's independent LLM sector

Knowledge Atlas Technology Joint Stock Company Limited (02513.HK; commercially known as Zhipu AI) is a leading player in China's independent large language model (LLM) sector (recognized for its capabilities and market position by Frost & Sullivan). The company has released its next-generation flagship model, GLM-5.2, achieving open-source State-of-the-Art (SOTA) performance in coding and agentic capabilities. It has also open-sourced the multimodal image generation model GLM-Image in collaboration with Huawei. Focusing on novel model architecture design, generalized reinforcement learning paradigms, and autonomous model evolution, its business layout closely aligns with the trend of enterprise-level AI productivity transformation.

MINIMAX-W

00100.HK

A top-tier player dedicated to omni-modal Artificial General Intelligence (AGI)

The company continuously iterates its video generation model matrix and has partnered with mainstream domestic and overseas chip manufacturers and inference platforms to complete underlying adaptations. The total monthly active users (MAU) of its B2C and B2B LLM applications, along with multimodal API call volumes, maintain rapid growth, further broadening its computing power ecological layout.

MININGLAMP-W

02718.HK

LLM concept stock, dubbed the "First Agentic AI Stock in Hong Kong"

The company fully underpins its Agentic AI positioning with its self-developed edge models—Mano, Cito, and Mano-P—alongside the Cider inference framework. Its business architecture utilizes the DeepMiner LLM as the underlying engine and Octo as the collaborative hub, uniformly delivering various AI products and industry solutions in the form of Agentic Services. Mininglamp pursues a differentiated technological route, bypassing the parameter expansion of general models in favor of a "Scaling Out" approach that coordinates multiple specialized small models. Its core moats do not rely on parameter scale, but rather on granular scenario data, specialized models, and continuous learning. This enables it to achieve a level of precision in vertical scenarios that general models cannot match, as it remains committed to building an open-source, privately deployable, and white-box auditable Private AI infrastructure.

KUAISHOU-W

01024.HK

A leading multimodal content generation platform in China

The company plans to restructure Kling AI and introduce external financing. Kling AI's quarterly revenue has reached 6.5 millions of RMB, with an Annualized Run Rate (ARR) exceeding USD 500 million. Its general LLM "KwaiYii" and video generation model "Kling" are undergoing continuous iteration, demonstrating high commercial conversion efficiency in scenarios such as e-commerce live streaming and automated marketing content generation.

MEITUAN-W

03690.HK

An AI technology pioneer in the local lifestyle services sector.

Its AI-native community "Miyou" has validated its practical value in physical operational scenarios, helping catering brands improve operational efficiency. The company continues to optimize search, recommendation, and dispatch algorithms within the local lifestyle ecosystem using its self-developed LLM capabilities. By introducing AI agent technologies into intelligent customer service, merchant marketing assistants, and on-demand delivery route planning, it remains committed to enhancing bilateral operational efficiency and refined management levels. 

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.

×
Share to WeChat

Open WeChat, use the "Scan", and share to my Moments.