Recent shifts in the Hong Kong stock market signal a series of positive developments. First, the Hang Seng Tech Index has refined its compilation rules, expanding constituent stocks and incorporating revenue growth into its screening criteria, directly addressing market concerns over the index's insufficient "tech content." Second, from the perspective of the latest industry trends, AI industry profits are shifting from AI hardware and large language models toward cloud providers, benefiting Hong Kong-listed companies. Looking ahead, we maintain a constructive outlook on investment opportunities in the Hong Kong market.
Hang Seng Tech Index Expansion
(1) This revision first reclassifies the technology theme system, marking the most structurally significant change of the reform. The proposed plan eliminates the original industry classification requirements and redefines tech companies into six major themes: internet platforms and solutions, artificial intelligence, advanced hardware, robotics and automation, cloud, and frontier technology. (2) Another core change is increasing constituent stocks from 30 to 50, introducing a dual-group stock selection mechanism. The selection universe is limited to Hang Seng Composite LargeCap and MidCap Index constituents; the selection method shifts from pure market cap ranking to a dual-track system of 40 stocks by market cap plus 10 stocks by revenue growth; and the total constituent count rises from 30 to 50. (3) Post-revision, the index exhibits stronger growth characteristics. According to Hang Seng Indexes Company's simulation using June 30, 2026 as the data cutoff, the 10 new "revenue growth group" constituents in the proposed Hang Seng Tech Index have a median revenue growth rate of 82.0% over the past 12 months, significantly higher than the 23.4% for the 10 new market cap group constituents and the 13.8% for the existing 30 constituents.
Industry Trend Shifts Favor Cloud Providers
(1) The current market trading logic centers on AI profits migrating from hardware to cloud providers, with Hong Kong stocks dominated by "soft tech" companies like cloud vendors benefiting the most. Within the AI industry chain's value distribution, industrial profits are undergoing a new round of redistribution, with cloud provider profit margins improving. Furthermore, the recent launch of high-performance open-source models such as Kimi 3 has reinforced this trend. These models continue to approach or even challenge certain closed-source models in complex reasoning and code generation tasks, reducing developers' and enterprises' reliance on a single closed-source model ecosystem and weakening the premium pricing power of leading model providers based on performance differentials. (2) Recent earnings reports from North American cloud providers validate accelerated cloud revenue growth, providing industrial mapping for Hong Kong cloud providers such as Alibaba and Tencent. The four major North American cloud providers reported combined quarterly cloud revenue of approximately $116.2 billion, up about 43% year-over-year. More critically, their combined remaining performance obligations reached approximately $2.33 trillion, surging 188% year-over-year, offering exceptionally high visibility for revenue growth over the next 2-3 years. This round of US cloud provider earnings reports reveals common characteristics: comprehensively accelerated cloud revenue growth, order backlogs (RPO/Backlog) growing well beyond market expectations, and customer demand expanding from a handful of frontier labs to a broad base of enterprise clients.
Risk Warning: Economic data falling short of expectations, incomplete policy understanding, and overseas policy tightening exceeding expectations.
Hang Seng Tech Index Expansion Details
On August 10, Hang Seng Indexes Company issued a consultation paper seeking market feedback on proposed amendments to the Hang Seng Tech Index compilation methodology. This marks the fifth compilation revision since the index's launch in 2020, with the final plan expected to be announced by end-September and officially implemented during the December index adjustment.
1. Technology Theme Scope Significantly Expanded, AI Upgraded to a Primary Theme
This revision first reclassifies the technology theme system, representing the most framework-defining change of the reform. The current Hang Seng Tech Index primarily covers internet, e-commerce, digital and intelligent solutions. However, as Hong Kong's tech industry structure continues to expand, traditional internet can no longer fully represent the Hong Kong tech sector. The proposed plan removes the original industry classification requirements, redefining tech companies into six major themes: internet platforms and solutions, artificial intelligence, advanced hardware, robotics and automation, cloud, and frontier technology. It also expands tech sub-themes from 16 to 24. Most notably, artificial intelligence is elevated from a sub-theme under the original "intelligence" category to one of the six primary tech themes, with new sub-directions such as "AI infrastructure" added. Additionally, frontier technology, as a newly introduced theme, will cover emerging fields including aerospace and satellite technology, quantum computing, and brain-computer interfaces. Post-revision, advanced hardware increases from 5 to 15 constituents, internet platforms' weight declines from 46% to 28%, and AI rises from 3 to 6 constituents, with a notably more balanced theme distribution.
2. Constituent Expansion with New Revenue Growth Screening Mechanism
Another core change in this revision is increasing constituent stocks from 30 to 50 while introducing a dual-group stock selection mechanism. The selection universe is limited to Hang Seng Composite LargeCap and MidCap Index constituents; the selection method shifts from pure market cap ranking to a dual-track approach of 40 stocks in the market cap group plus 10 stocks in the revenue growth group; constituent numbers rise from 30 to 50; and buffer zones are differentiated (market cap group ±20%, revenue growth group ±50%). Specifically, eligible companies are first ranked by market cap to select 40 stocks for the market cap group. The remaining eligible companies not in the top 40 by market cap are then ranked by revenue growth rate over the past 12 months, with the top 10 forming the revenue growth group. Together, the two groups comprise 50 constituents. Compared to the original market-cap-centric selection framework, the new mechanism incorporates growth considerations, preventing fast-growing tech companies with relatively small current market caps from being excluded due to insufficient scale. Consequently, the Hang Seng Tech Index's style shifts from purely favoring large-cap tech toward balancing large-cap leaders with high-growth tech companies. Introducing the revenue growth group is the most substantive innovation of this reform. The Hong Kong tech sector has long faced a pain point: numerous hard-tech companies in high-growth phases, such as AI chip and robotics firms, have been persistently excluded from the index due to lack of profitability and small market caps. The revenue growth group effectively provides a fast-track channel for such companies.
3. Post-Revision Index Exhibits Stronger Growth Characteristics
According to Hang Seng Indexes Company's simulation using June 30, 2026 as the data cutoff, the 10 proposed new "revenue growth group" constituents in the Hang Seng Tech Index demonstrate a median revenue growth rate of 82.0% over the past 12 months, markedly higher than the 23.4% for the 10 new market cap group constituents and the 13.8% for the existing 30 constituents. This indicates that the revenue growth group significantly enhances the index's coverage of small-cap, high-growth tech companies. However, the group's simulated aggregate weight is only 2.4%, so its impact on the index's overall financial characteristics and performance returns requires further observation.
AI Industry Profit Shifts from Hardware to Cloud Providers
From the perspective of the latest industry trends, the total profit pool in the AI industry has not been revised upward, yet the distribution structure is shifting from AI hardware and large language models toward cloud providers. Hong Kong, home to cloud providers like Tencent and Alibaba that are scarce in the A-share market, has recently demonstrated relatively strong performance.
1. Profit Migration from Hardware to Cloud Providers
The recent market trading logic centers on AI profits shifting from hardware to cloud providers, with Hong Kong stocks dominated by "soft tech" such as cloud companies benefiting the most. Within the AI industry chain's value distribution, industrial profits are experiencing a new round of redistribution, with cloud provider profit margins improving. In the early phase, closed-source large models commanded strong pricing power through leading model performance and technical barriers. Capital expenditure primarily revolved around model training, driving demand for infrastructure such as GPUs and servers, making hardware manufacturers the most direct beneficiaries in the AI industry chain. However, as model capabilities rapidly iterate, open-source ecosystems continue developing, and inference costs gradually decline, the premium pricing power of closed-source models diminishes (evidenced by recent OpenAI price cuts and Anthropic's second derivative of ARR turning negative). The AI value chain is shifting from concentrated benefits in a single segment toward a more balanced distribution pattern. On one hand, cloud providers, leveraging advantages in computing infrastructure, model deployment platforms, and AI application entry points, are positioned to enhance value capture in model invocation and enterprise services. On the other hand, consumers will also share in the benefits of technological progress through the proliferation of AI applications and productivity gains. The recent launch of high-performance open-source models such as Kimi 3 further reinforces this trend. Their capabilities in complex reasoning and code generation continue approaching or even challenging certain closed-source models, reducing developers' and enterprises' dependence on a single closed-source model ecosystem and weakening the premium pricing power of leading model providers based on performance differentials. As open-source model performance improves and commercialization paths mature, future AI industry competition may shift from model capability competition to computing efficiency, application ecosystem, and user scale competition. Industrial profits are expected to be further redistributed toward cloud infrastructure, application services, and end-user segments. The latest North American earnings reports also validate this trend, with cloud provider profitability continuing to rise. The market previously worried that AI capital expenditure would suppress cloud provider profit margins, but earnings data have disproved this pessimistic expectation. AWS and GCP operating margins reached approximately 39% and 36% respectively, while Microsoft Intelligent Cloud maintained an operating margin of around 41%. Improved computing utilization rates, differentiated pricing, and software-hardware synergy optimization are offsetting rising depreciation and operational costs.
2. Overseas Cloud Business Continues Accelerating
Recent North American cloud provider earnings reports validate accelerating cloud revenue growth, providing industrial mapping for Hong Kong cloud providers such as Alibaba and Tencent. The four major North American cloud providers reported combined quarterly cloud revenue of approximately $116.2 billion, up about 43% year-over-year. More critically, their combined remaining performance obligations reached approximately $2.33 trillion, surging 188% year-over-year, offering exceptionally high visibility for revenue growth over the next 2-3 years. The simultaneous upward movement of revenue, profit margins, and backlog orders indicates that earlier capital expenditure is rapidly converting into current revenue, operating profit, and future revenue visibility. This round of US cloud provider earnings reports reveals common characteristics: comprehensively accelerated cloud revenue growth, order backlogs (RPO/Backlog) growing well beyond market expectations, and customer demand expanding from a handful of frontier labs to a broad base of enterprise clients. In the latest quarter, Azure/AWS/GCP revenue grew 43%/37%/82% year-over-year respectively, all showing clear acceleration. Microsoft's commercial RPO reached $678 billion, up 84% year-over-year, with all new sequential orders coming from non-frontier model customers, indicating demand is spreading from a few AI labs to broad enterprise clients. AWS's backlog reached approximately $496 billion, more than doubling year-over-year, with management expecting most newly added computing capacity for FY27 to be pre-booked by customers. GCP's backlog reached $514 billion, increasing by more than $50 billion sequentially, with AI solutions having become a significant driver of Cloud growth.
Comments