MiniMax and Zhipu join Hang Seng Tech Index

๐กFirst-ever inclusion of pure-play AI firms in a major benchmark index, signaling a shift in AI market maturity.
โก 30-Second TL;DR
What Changed
MiniMax and Zhipu are the first pure-play AI firms in the index
Why It Matters
The inclusion signals growing institutional recognition of AI as a core sector in the Chinese tech economy, potentially increasing investor scrutiny of these firms.
What To Do Next
Track the financial disclosures and model performance reports of MiniMax and Zhipu to better understand the commercial viability of Chinese LLMs.
Key Points
- โขMiniMax and Zhipu are the first pure-play AI firms in the index
- โขInclusion expected to drive passive capital inflows
- โขMarket performance was mixed following the announcement
๐ง Deep Insight
Web-grounded analysis with 23 cited sources.
๐ Enhanced Key Takeaways
- โขThe inclusion of MiniMax Group and Zhipu AI in the Hang Seng Tech Index was officially announced on May 22, 2026, and became effective on June 8, 2026.
- โขThis move is projected to generate substantial passive capital inflows, estimated to be between $1.25 billion and $1.8 billion, into these AI firms.
- โขThe addition of these pure-play AI companies is part of a broader strategy by Hang Seng Indexes Co. to expand the index to 100 constituents and address prior criticism that the index had underperformed by not adequately capturing the global artificial intelligence boom.
- โขBoth MiniMax and Zhipu AI completed their Initial Public Offerings (IPOs) on the Hong Kong Stock Exchange in January 2026.
- โขFollowing its listing, Zhipu AI's shares surged by nearly 1,600% by May 2026, pushing its market capitalization above $112 billion, while MiniMax's shares doubled on its debut, briefly exceeding a $13 billion market value.
๐ Competitor Analysisโธ Show
| Company | Primary Focus | Key Models/Products | Market Position/Valuation (approx.) |
|---|---|---|---|
| MiniMax | Multimodal AI models, consumer AI apps, enterprise APIs | MiniMax-M2, Hailuo AI, Talkie, ABAB 6.5, Mavis AI Agent | One of China's "AI tigers"; $16B - $66B valuation range (March 2026) |
| Zhipu AI | General-purpose large models, enterprise LLM services, code generation | GLM family (GLM-4, GLM-5, ChatGLM), Ying (text-to-video), AutoGLM Rumination | Leading Chinese independent LLM developer; $112B market cap (May 2026) |
| Baidu | Integrated AI across search, cloud, autonomous driving | Ernie Bot | Major incumbent, leveraging search data and cloud scale |
| Alibaba | Cloud AI, enterprise solutions, e-commerce integration | Qwen | Major incumbent, leveraging cloud infrastructure |
| Moonshot AI | Long-context large language models | Kimi | Key "AI tiger" startup, known for natural conversations |
| SenseTime | Computer vision, broader AI solutions | SenseChat | Established AI company, also developing LLMs |
๐ ๏ธ Technical Deep Dive
- Zhipu AI:
- Developed the GLM (General Language Model) family of large language models.
- Launched the GLM framework in 2021, which was China's first proprietary pre-trained large model framework.
- Open-sourced GLM-130B in 2022, marking China's first 100-billion-parameter large language model.
- Features the GLM-4 series, including GLM-4.5, a 355-billion-parameter open-source foundation model designed for advanced reasoning, agentic behavior, coding tasks, and multimodal capabilities.
- Released the GLM-5 series in 2026, with performance claimed to be comparable to models like GPT-5 and Anthropic's Claude Opus 4.5.
- Introduced GLM-5.1 in April 2026, which demonstrated breakthroughs in response speed and enables agents to build sophisticated websites rapidly.
- Debuted the Ying text-to-video model in July 2024 and GLM-4-Voice, an end-to-end speech LLM capable of adjusting tone and dialect, in October 2024.
- Launched AutoGLM Rumination in March 2025, a free AI agent claiming to be eight times faster than DeepSeek's R1.
- GLM 4.5 (July 2025) is capable of running on eight NVIDIA H20 chips, and GLM-4.6 (September 2025) integrated FP8 and Int4 quantization on Cambricon chips, also supporting native FP8 on Moore Threads GPUs.
- MiniMax:
- Develops multimodal AI models capable of processing text, audio, images, video, and music.
- Offers consumer applications such as the AI character apps Talkie and Xingye, and the video-generation service Hailuo AI.
- Provides enterprise API services for its models.
- Launched the ABAB 6.5 series, a mixture of experts language model, in April 2024.
- Unveiled the MiniMax-01 LLM product line in January 2025, which includes the general-purpose MiniMax-Text-01 and the visual-capable MiniMax-VL-01.
- Released Speech-02 in April 2025, a text-to-speech model supporting over 30 languages.
- Hailuo AI, launched in March 2024, is an AI text and music-generating LLM platform, which later included the Video-01 text-to-video model in September 2024.
- Its proprietary foundation model suite, including MiniMax-M2, Hailuo-02, and Speech-02, features long context processing capacity.
- Showcased M3 Sparse Attention (MSA) technology and released a technical report on its M2 series in May 2026, highlighting improvements in long-context processing and agent training systems.
- Upgraded its AI Agent product to "Mavis" in May 2026, introducing an "Agent Team" collaborative mode to enhance practical implementation.
๐ฎ Future ImplicationsAI analysis grounded in cited sources
โณ Timeline
๐ Sources (23)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
Weekly AI Recap
Read this week's curated digest of top AI events โ
๐Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: SCMP Technology โ
