HKEX Launches AI-Focused Index Comprising 40 Hong Kong-Listed AI Companies
nashnova research
HKEX has launched a dedicated AI index selecting 40 Hong Kong-listed companies across chips, infrastructure, models, and applications — with screening rules that ignore profitability and reward R&D spending and revenue growth, laying the groundwork for AI-themed ETFs.
Who made the cut, and how are they grouped?
The 40 constituents span four AI themes: AI chips, AI infrastructure, AI models, and AI applications.
The list includes mega-caps like Alibaba, Tencent, and Baidu alongside recent Hong Kong IPOs such as Biren Technology, Zhipu, and MiniMax.
This means → the index deliberately places loss-making AI newcomers in the same basket as mature tech giants — this is not a blue-chip rebrand.
If not profitability, what does the scoring measure?
HKEX's fundamental score rests on three metrics only: R&D spending as a share of revenue, capex as a share of revenue, and year-on-year revenue growth. It uses two full fiscal years of data, weighting the most recent year at 60%.
Revenue growth carries a scoring coefficient of 2 — double that of R&D intensity or capex intensity. Growth speed matters most.
In plain terms = the rules ask two questions: "How much are you spending on R&D?" and "How fast is your revenue growing?" Whether you turn a profit is, for now, irrelevant.
This reflects HKEX's acceptance of the AI "burn cash for scale" phase: Biren's H1 revenue surged 1,997.6% yet it still lost RMB 377 million; Zhipu lost RMB 2.07 billion; MiniMax lost USD 358 million. Under traditional index rules, none would qualify.
Which end of the value chain gets the most weight?
Weight caps differ by theme: a single AI-chip, AI-infrastructure, or AI-model stock can weigh up to 10%; an AI-application stock is capped at just 3%.
As of October 22, the top three by weight were SMIC at 9.8%, Hua Hong Semiconductor at 8.26%, and Lenovo at 6.76%. Alibaba, Tencent, and Baidu each sat in the 3.5%–3.8% range.
This means → the index tilts toward "build the chips, build the compute" hard-tech companies. Internet platforms made the list but carry roughly the same weight as application-layer firms.
What are the entry requirements and rebalancing rules?
Key thresholds: at least 30% of revenue from one or more AI themes; non-constituents generally need a minimum 3-month listing history; average daily turnover of at least HKD 20 million over the past six months.
The index rebalances quarterly — in March, June, September, and December. HKEX says newly listed AI companies that meet the criteria can enter through regular reviews.
In plain terms = the deck is reshuffled every three months — a qualifying AI newcomer can join in as little as one quarter.
How large is HKEX's index empire now?
Since launching its first proprietary index — the HKEX Tech 100 — in December 2025, HKEX has rolled out at least seven proprietary or co-branded equity indices, spanning cross-market products (HK-Malaysia, HK-Korea, HK-US) and thematic plays (AI, semiconductors).
The IPO backdrop: from December 2025 through May 2026, AI value-chain companies raised HKD 97.9 billion in Hong Kong IPOs, accounting for roughly 55% of total new-listing proceeds.
This reflects HKEX linking "attract AI companies to list" with "build indices and launch ETFs around them" into a single product chain.
When can investors buy an ETF tracking this index?
HKEX has signed a licensing agreement with China Southern Asset Management to launch a mainland-distributed ETF tracking the AI index.
E Fund (Hong Kong) already listed the first ETF tracking the Tech 100 index in June; four more ETFs tracking HKEX cross-market indices launched in September.
This means → the index itself is just a yardstick; the real commercial loop is ETFs. Whether it attracts meaningful tracking capital is the test of market acceptance. Hang Seng Indexes Company launched a similar 40-constituent AI index back in 2023 — the two differ in stock selection and weighting, and the market will ultimately vote with its money.
市场有风险，内容仅供研究参考，不构成投资建议。