After a lackluster end to the second quarter, the Hang Seng Tech Index has re-entered the focus of capital markets. On July 27, 2026, the index hit an intraday high of 4,726.3 points, a notable rebound from its recent low of 4,229.94 points on June 26, though it remains roughly 30% below its peak of 6,715.46 points from October 2, 2025.

Southbound capital has concurrently returned. Public data shows that from July 1 to 20, 2026, cumulative net purchases of Hong Kong stocks amounted to approximately HK$33.65 billion, bringing the year-to-date total net buying to around HK$379.644 billion.

The most straightforward explanation for this recovery is the convergence of price action, capital inflows, and low valuations. However, the return of capital does not equate to a fundamental shift in the earnings logic. The current rally in the Hang Seng Tech Index is primarily a valuation repair. AI-driven earnings revaluations are occurring only at a select few companies and are insufficient to support a complete valuation system overhaul for the entire index. The excessive discount previously assigned by the market and the correction of overly pessimistic expectations can fuel the initial upward leg. Whether the index can sustain its ascent depends not on the number of new model releases, but on whether AI-generated revenue, profit increments, and free cash flow can cover the associated increases in capital expenditure, depreciation, and research and development spending.


Valuation Repair Partially Complete; Index is Not Universally Cheap

On May 27, 2026, the Hang Seng Tech Index's trailing twelve-month (TTM) P/E ratio was approximately 20.14x, placing it in a low 9.31st percentile over the past five years. By July 20, the P/E had recovered to about 22.61x, lifting its five-year percentile to roughly 37.31%, with a price-to-book ratio of around 2.5x. While these two data points are from different periods, they effectively illustrate the valuation shift: the market was indeed in a cheap zone near the end of June, but with the July rebound, the easiest "mean reversion" gains have already been partially realized. A P/E of around 22x is not expensive, but it can no longer be described as "extremely undervalued."

More importantly, the Hang Seng Tech Index is not a homogenous index of a single business model. Tencent Holdings Ltd (HKG: 0700) and Alibaba Group Holding Ltd (HKG: 9988) possess high-cash-flow platform businesses. Kuaishou Technology (HKG: 1024) relies on advertising, e-commerce, and generative video. Meituan (HKG: 3690) and JD.com Inc (HKG: 9618) are engaged in intense retail competition. Xiaomi Corp (HKG: 1810) spans smartphones, AIoT, internet services, and electric vehicles. Packaging these diverse companies into a single index P/E ratio can obscure differences in earnings quality, capital intensity, and growth drivers.

The persistent discount of the Hang Seng Tech Index compared to major overseas tech stocks is not entirely a mispricing. It rationally incorporates factors such as heavier competitive investments, less stable shareholder returns, losses from new business ventures, and geopolitical risks. Therefore, the current valuation truly reflects a dual-layer expectation: on one hand, the market no longer believes the platform economy will continuously deteriorate, with buybacks, cash reserves, and business resilience forming a valuation floor. On the other hand, the market remains reluctant to pay a high premium for unproven AI narratives. The continuous inflow of Southbound capital proves that onshore funds are willing to absorb low-valuation assets, but it alone cannot prove that a structural upward revision of earnings forecasts has occurred. Capital flows are a catalyst; profit and cash flow are the basis for revaluation.


AI Commercialization is Emerging, but Not Yet Covering All Heavyweights

Tencent represents a "strong core business efficiency improvement with simultaneous AI investment growth" scenario. In the first quarter of 2026, the company reported revenue of RMB 196.458 billion, a 9% year-over-year increase. Non-IFRS attributable net profit was RMB 67.905 billion, up 11% year-over-year. Marketing services revenue reached RMB 38.171 billion, up 20% year-over-year, while fintech and business services revenue was RMB 59.885 billion, a 9% increase, with enterprise services revenue growing 20% and Tencent Cloud's international business revenue surging over 40%. The company disclosed that AI M+ has covered about 30% of advertisers' budgets. Excluding investments in new AI products, non-IFRS operating profit growth would have been 17% year-over-year. This demonstrates that AI is genuinely improving ad conversion and cloud business, but new investments are also consuming some of the profits. Tencent has the capacity to use cash flow to support AI spending, but it still needs to prove that these investments will not persistently pressure profit growth and buyback intensity.

Alibaba's evidence of AI commercialization is more direct, but the cost is also clearer. For the quarter ending March 31, 2026 (FY2026 Q4), the company reported revenue of RMB 243.4 billion, a 3% increase year-over-year, or 11% on a like-for-like basis excluding divested businesses. Cloud intelligence revenue was RMB 41.6 billion, up 38% year-over-year, with external cloud commercialization revenue growing 40%. AI-related product revenue was approximately RMB 8.971 billion, accounting for about 30% of external cloud commercialization revenue, and has achieved triple-digit growth for the 11th consecutive quarter. The challenge lies in the fact that investments in instant retail, Qianwen (Tongyi Qianwen) customer acquisition, and cloud infrastructure caused adjusted EBITA to fall to approximately RMB 5.1 billion, a decline of 84% year-over-year, with free cash flow recording a net outflow of RMB 17.3 billion. Alibaba's AI revenue is no longer a concept, but whether the group's valuation can be re-rated depends on whether cloud business growth can translate into more stable profits and cash returns, rather than just revenue growth rates.

Kuaishou is another company that has produced direct revenue evidence. In the first quarter of 2026, the company reported revenue of RMB 33.716 billion, a 3.4% increase year-over-year, but adjusted net profit attributable to shareholders was RMB 3.374 billion, a 26.3% decline. Advertising revenue was RMB 19.643 billion, up 9.3% year-over-year. AI recommendations and intelligent bidding lifted domestic marketing revenue by approximately 3% to 4%, while OneSearch V2 boosted e-commerce search GMV by about 3%. Kling AI generated quarterly revenue exceeding RMB 650 million, a year-over-year increase of over 300%, and its annualized recurring revenue (ARR) was close to US$500 million as of March 2026. Unlike simple efficiency tools, Kling has formed an independent paid revenue stream. However, the company's 2026 capital expenditure guidance is approximately RMB 26 billion, with core business growth slowing, profits declining, and high investment continuing. Therefore, Kuaishou is closer to a scenario where its "AI business has been revalued" rather than the entire company's profits entering a period of effortless expansion.

JD.com, Meituan, and Xiaomi illustrate that AI is not yet the primary pricing variable for all heavyweight stocks. In the first quarter of 2026, JD.com reported revenue of RMB 315.7 billion, a 4.9% year-over-year increase, and non-IFRS net profit of RMB 7.4 billion, a decline of about 42% year-over-year. JD Retail's operating profit margin was 5.6%, while new businesses reported an operating loss of RMB 10.4 billion. While AI improvements in advertising efficiency are noteworthy, the group's forecast P/E ratio of around single digits mainly reflects concerns over low retail growth, investments in instant retail, and losses from new ventures.

Meituan reported first-quarter 2026 revenue of RMB 91.039 billion, a 5.6% year-over-year increase, but recorded an operating loss of RMB 6.47 billion and an adjusted net loss of RMB 4.968 billion. Core local commerce posted an operating loss of RMB 2.03 billion, while new ventures had an operating loss of RMB 2.116 billion. The company's AI tools are already serving a large number of merchants, with "Smart Manager" covering over 700,000 food and beverage merchants and "Digital Employee" serving over 300,000 service retail merchants. However, at this stage, these tools are primarily used to improve merchant and platform efficiency. The primary issue determining Meituan's valuation remains the intense competition in instant delivery, subsidy levels, and when the unit economics will recover, rather than whether AI revenue alone can form a valuation anchor.

Xiaomi reported first-quarter 2026 revenue of RMB 99.142 billion, a 10.9% year-over-year decline, and adjusted net profit of RMB 6.072 billion, a 43.1% drop. Revenue from smartphones and AIoT was RMB 79.3 billion, while revenue from smart EVs and AI innovation businesses was RMB 19.9 billion. R&D spending increased 33.4% year-over-year to RMB 9 billion. Xiaomi's progress in large models and on-device AI can strengthen its "Human x Car x Home" ecosystem, but short-term profits are still primarily influenced by memory and commodity price hikes, smartphone competition, and EV business investments. A more reasonable framework for Xiaomi is a sum-of-the-parts valuation for its consumer electronics, internet services, automotive, and AI innovation businesses, rather than simply treating it as an internet AI company.


Moving from Valuation Repair to Earnings Revaluation Requires Three Financial Statements

Whether the Hang Seng Tech Index can complete the transition from valuation repair to earnings revaluation will ultimately need to be verified simultaneously across the income statement, cash flow statement, and capital expenditure budget. First, AI revenue needs to evolve from a "high-growth, small base" into a business segment significant enough to influence the group's revenue structure. Second, efficiency gains in advertising, cloud services, and subscription revenue must translate into operating profit, rather than being entirely offset by computing power, depreciation, and customer acquisition costs. Third, free cash flow cannot be sustained by merely cutting buybacks or drawing down the balance sheet. If revenue growth is achieved while profits and cash flow continue to weaken, the market is more likely to grant a thematic premium rather than a stable earnings revaluation.

By this standard, Alibaba and Kuaishou have already crossed the first threshold of direct AI revenue. Tencent has strong evidence of efficiency gains in advertising and cloud services. However, all three companies still need to answer the question of return on investment. Meituan, JD.com, and Xiaomi have not yet entered a phase where AI profits are the primary valuation anchor. Their stock prices are more susceptible to competitive dynamics, supply chain costs, EV deliveries, and losses from new businesses.

Consequently, the Hang Seng Tech Index will continue to diverge internally. An index rally does not mean all component stocks are simultaneously undergoing an AI revaluation. This is where the current market is most prone to confusion: a low-valuation bounce can be quickly achieved by capital flow and risk appetite, but an earnings revaluation requires multiple quarters of data accumulation. The recovery of the Hang Seng Tech Index's P/E from about 20x to over 22x has already proven that a portion of the previous discount was overly pessimistic. However, without continuous upward revisions to earnings forecasts, further valuation expansion will increasingly depend on liquidity. What has not been fully priced in is not "Hong Kong tech companies also have AI," but which specific company can use AI to improve its return on capital while maintaining free cash flow during a high-investment cycle.

Therefore, a more accurate definition of this Hang Seng Tech rally is "valuation repair has begun, and earnings revaluation is occurring in pockets." The cheapness at the index level is diminishing, while the divergence at the individual stock level is increasing. The focus for subsequent judgment should not be on whether Southbound capital continues to flow in, nor should it be limited to ranking model capabilities. Instead, investors should track the changes in the proportion of cloud and AI revenue, advertising efficiency, capital expenditure, depreciation, operating profit, and free cash flow. Only when these metrics form a positive feedback loop will AI truly transition from a valuation narrative into an earnings report.