Hang Seng TECH: As old headwinds fade, new earnings drivers emerge

• Earnings downgrades are becoming less broad-based after a substantial reset. Post-2Q revisions have become more balanced: hardware leaders and selected platforms saw upgrades or stable forecasts, while downgrades were increasingly concentrated in companies with heavy AI spending.

• Hardware now provides greater earnings support to Hang Seng TECH Index (HSTECH). SMIC, Lenovo and Hua Hong account for an increasing weight of the index. AI-infrastructure demand, higher foundry pricing and strong utilisation are driving upward earnings revisions across these companies.

• Platform headwinds are easing, while AI returns are emerging gradually. Food-delivery competition has moderated and Alibaba’s cloud investment is generating visible revenue and margin improvement. Tencent has demonstrated AI-driven gains in advertising and cloud, although 2027 is likely to remain an investment and validation year.

• Low valuation lowers the bar for a rebound. With HSTECH trading near historical valuation lows, a broad earnings recovery may not be required to trigger a re-rating. Stabilising earnings estimates, continued hardware upgrades or clearer evidence of AI monetisation could be sufficient catalysts.

• Prospective index changes should improve HSTECH’s technology representation. Once implemented, the revised methodology should broaden exposure to advanced hardware, AI and digital infrastructure, while its revenue-growth criterion could give smaller, faster-growing technology companies a better opportunity to enter the index.

Since mid-2025, the Hang Seng TECH Index (HSTECH) has faced several simultaneous headwinds, prompting multiple rounds of substantial consensus earnings downgrades. In our view, this reset has already incorporated a meaningful portion of the negative news.

Following the second-quarter 2026 results season, earnings estimates were raised or kept broadly unchanged for companies representing around 60% of the index's top-ten weight. Earnings expectations stabilised or improved for food-delivery competitors such as Meituan and JD.com, while hardware companies—most notably SMIC and Lenovo—received substantial upgrades. These positive revisions were partly offset by further downgrades concentrated in a small number of large internet platforms, particularly Alibaba and Baidu.

The second quarter therefore did not mark the beginning of a broad-based earnings recovery. Rather, it showed that expectations have stabilised across much of the index, while earnings trends have become increasingly differentiated at the company level.

At this point, some of the index's earlier risks are beginning to ease, while new sources of earnings growth are becoming more visible. We therefore maintain a positive view on the iShares Hang Seng TECH ETF (HKEX: 3067).

Hardware is becoming more important to HSTECH

SMIC, Lenovo and Hua Hong Semiconductor together account for approximately 16.2% of HSTECH, giving their earnings revisions an increasingly meaningful influence on index-level earnings.

SMIC is mainland China’s largest and most technologically advanced foundry, with operations spanning advanced logic, mature nodes and specialty processes. Over the past two years, weak consumer-electronics demand, capacity expansion and falling prices have pressured the mature-node industry. SMIC’s elevated capital expenditure and depreciation from new capacity have created an additional margin drag, keeping market expectations subdued.

This backdrop is now starting to shift, as AI-related demand tightens supply across mature and specialty processes.

AI demand is tightening supply beyond leading-edge chips

The current improvement is being driven by two related factors.

First, rising global AI data-centre investment is increasing demand not only for leading-edge processors but also for supporting chips such as power-management ICs, BCD power-management chips, connectivity chips and optical-communications devices. Although these products are generally manufactured using mature or specialty processes, the relevant capacity is not easily interchangeable. Suppliers must offer stable yields and process-specific capabilities, while lengthy customer-qualification cycles make it difficult to switch foundries at short notice. Stronger AI-related demand is therefore tightening supply in selected mature and specialty processes, giving qualified foundries greater pricing power.

According to TrendForce, average mature-node foundry prices increased by approximately 5%–15% during the first two quarters of 2026 as global supply-demand conditions tightened. This has allowed SMIC to raise prices selectively where it has process advantages and customers cannot easily switch suppliers. Its blended average selling price rose approximately 5.7% quarter on quarter in the second quarter, with some new contract prices expected to become more visible in the third quarter.

Second, overseas foundries are allocating more capacity to higher-margin AI processors, memory and high-bandwidth products. Even within mature nodes, they are prioritising supporting chips used in AI infrastructure. This has constrained capacity for conventional mature-node products and redirected some orders—including image sensors and display-driver ICs—to Chinese foundries. SMIC and Hua Hong, which offer stable production and additional capacity, are natural beneficiaries.

SMIC’s results indicate that demand is absorbing its capacity additions. Despite new capacity coming online in both the first and second quarters, utilisation remained high at 93.1% and 93.7%, respectively. Management also indicated that orders for certain BCD power-management products are visible through the end of 2027, suggesting that tight supply for AI-related supporting chips could persist.

SMIC expects third-quarter revenue to rise another 2%–4% sequentially and gross margin to reach 26%–28%, well above the approximately 22% expected before the results. High utilisation, higher contract prices and an improving product mix should help offset depreciation from new capacity.

In the near term, SMIC’s earnings recovery is being driven mainly by tighter supply-demand conditions in mature and specialty processes. Over the longer term, advanced-node manufacturing offers additional strategic value. SMIC is the only mainland Chinese foundry with relatively established experience in the mass production of seven-nanometre-class chips, making it a critical platform for domestic chip designers seeking local advanced manufacturing.

The recovery is spreading beyond SMIC

Hua Hong’s performance suggests that the improvement is spreading across the broader mature-node and specialty-process market. Second-quarter revenue increased 26.8% year on year to a record USD 718 million, while gross margin rose from 10.9% to 16.5%. Better pricing, high utilisation, cost controls and stronger demand for specialty products such as non-volatile memory supported the improvement. The company expects third-quarter revenue of USD770–780 million and a gross margin of 16%–18%, indicating that favourable conditions are continuing in selected segments.

The same demand chain is evident at the server level. Lenovo’s Infrastructure Solutions Group recorded 98% revenue growth in the latest quarter, while operating margin rose from 3.6% to 9.1%. Its AI-server sales pipeline reached USD 54 billion. The customers require system integration, cooling, maintenance and financing services, allowing Lenovo to generate additional revenue beyond server hardware.

Policy support further reinforces this domestic supply chain. China’s third National Integrated Circuit Industry Investment Fund is supporting the localisation of wafer manufacturing, equipment and materials, benefiting SMIC and Hua Hong through capacity construction, process validation and domestic tape-out demand. Meanwhile, investment in AI and digital infrastructure is driving server and computing-equipment demand, supporting Lenovo’s downstream orders. Together, these developments are allowing more of the value created by AI hardware investment to remain within China’s domestic supply chain.

Platform competition is easing, while returns on AI investment are becoming partly measurable

The turning point in the food-delivery subsidy war is becoming clearer. Since the start of 2026, growth in selling and marketing expenses—a proxy for platform subsidies—has slowed across the three major participants. China’s State Administration for Market Regulation has also signalled a firmer stance against prolonged, large-scale subsidies that disrupt competition. In the second quarter, Alibaba’s and JD.com’s selling and marketing expenses declined 10.4% and 24.8% year on year, respectively, while growth in Meituan’s marketing expenses slowed to 11.5%. Alibaba’s instant-commerce unit economics improved and its losses narrowed, while JD.com’s cash flow recovered. These developments suggest that the most aggressive phase of competition has passed, allowing platforms gradually to rebuild the profits and cash flow needed to fund AI investment.

Another source of earnings uncertainty has been whether substantial AI capital expenditure can generate visible commercial returns. Alibaba is further along because of its earlier investment and exposure across the AI value chain. In the latest quarter, total and external-customer revenue from its AI Cloud and computing-services segment both grew 45% year on year, the fastest pace in 22 quarters. The acceleration in external revenue indicates genuine enterprise demand rather than merely greater internal computing usage.

The improvement also extends beyond one quarter. Alibaba’s AI-related product revenue has grown at triple-digit rates for 12 consecutive quarters, reaching approximately RMB12.38 billion in the latest quarter. This represented around 35% of Alibaba Cloud’s external revenue and implied an annualised run rate of almost RMB49.5 billion. Adjusted EBITA for the segment rose 133% to RMB5.63 billion, while its margin increased from approximately 7.2% to 11.6%. Management has indicated that AI-related products carry higher gross margins than the broader cloud business.

Management estimates that the payback period for AI-related capital expenditure could shorten from the current three-year baseline to approximately 2.5 years as gross margins improve and proprietary-chip adoption increases. Together with soaring demand for cloud computing and AI services, this provides the clearest evidence yet that Alibaba’s AI investment is beginning to shift from a cash-flow burden into a measurable source of returns.

Tencent’s AI monetisation is progressing at two different speeds. AI enhancements to existing businesses are already delivering results, while AI-native products remain in the investment and validation stage. AI-powered recommendations and automated ad placement have improved matching, conversion rates and eCPM, supporting 22% growth in Marketing Services revenue. Tencent Cloud revenue growth also accelerated to the low-20% range, while GPU leasing, model services and enterprise AI tools began generating external demand. This demonstrates that AI can improve the monetisation efficiency of Tencent’s existing traffic and infrastructure.

However, Tencent has not yet entered a phase of broad AI-related profit contribution. It continues to invest in computing infrastructure, models and its application ecosystem, while the associated capital expenditure, depreciation and product-promotion expenses could remain elevated in 2027. Investment costs may therefore continue to constrain earnings growth and the recovery in free cash flow, even as AI-related revenue expands. The year is more likely to represent a transition in which AI revenue begins to offset part of the investment burden without yet making a significant contribution to group profit.

Tencent’s standalone AI products have nevertheless achieved encouraging early validation. WorkBuddy and CodeBuddy have recorded paid-user retention rates above 80% and active-user retention rates above 60%, suggesting initial product-market fit and user stickiness. Tencent can also use WeChat, WeCom, Tencent Meeting, Tencent Docs and Tencent Cloud to lower customer-acquisition costs. If WeChat’s “Xiaowei” personal assistant is successfully launched, these distribution and ecosystem advantages could gradually translate into AI-native revenue.

Proposed HSTECH expansion could broaden exposure to high-growth AI companies

When HSTECH was launched, Hong Kong's technology market was dominated by large internet and e-commerce platforms. China's technology sector has since broadened substantially towards advanced hardware, artificial intelligence and digital infrastructure.

Under the proposed methodology revisions—which have not yet taken effect—the index would expand its coverage of these areas, allowing it to reflect the evolving structure of China's technology industry more accurately and potentially increasing its relevance to global investors.

The proposed framework would also introduce a revenue-growth selection criterion. If implemented, this could shift HSTECH from a framework focused primarily on large, established technology leaders towards a more inclusive structure in which smaller but faster-growing companies have a better opportunity to qualify for inclusion.

Together, these proposed changes could broaden the index's exposure to companies participating directly in the fastest-growing areas of China's technology ecosystem. However, their impact should be treated as a prospective catalyst until the revised methodology is formally implemented and the resulting constituent changes become effective.

Key risks

• Macroeconomic risk: A sharper-than-expected deterioration in China's economy could weaken consumer spending, advertising budgets, cloud demand and technology investment, resulting in further earnings downgrades and delaying the index's valuation recovery.
• Re-escalation risk: If delivery-platform competition among Alibaba, Meituan and JD.com resumes rather than continuing to fade, renewed subsidies and marketing expenditure could keep platform margins under pressure and delay the 3067 recovery thesis.
• AI monetisation risk: AI-related revenue at Alibaba and Tencent remains a minority of group revenue and depends on continued customer adoption. Slower commercialisation or persistently elevated investment could weaken the earnings support required for a re-rating.
• Export-control risk: Tighter US export controls could further restrict SMIC's access to advanced semiconductor-manufacturing equipment, components and technical support. This could constrain capacity expansion, raise production costs and delay progress in advanced-node manufacturing.

Conclusion

HSTECH is moving from an earnings-downgrade cycle towards a period of earnings stabilisation and validation. Improving platform cash flow, gradual progress in AI commercialisation and the rising earnings contribution from hardware companies are providing greater fundamental support for the index.

The third quarter may lack a single major catalyst. However, with valuations near historical lows, stabilising earnings expectations could itself unlock meaningful rebound potential. Based on our updated assessment of constituent fundamentals, we estimate that HSTECH could still offer approximately 66% upside through 2028, supporting our positive view on the iShares Hang Seng TECH ETF (HKEX: 3067) as a vehicle for gaining exposure to this potential recovery.

Table 1: Hang Seng TECH Index earnings, valuation and target-price estimates

Declaration

For specific disclosure, at the time of publication of this report, IFPL, through its connected and associated entities, and the analyst who produced this report hold a NIL position in the abovementioned securities.

This research report was prepared with the assistance of artificial intelligence tools. iFAST Financial Pte Ltd does not rely exclusively on AI for content generation. The content of this report, including all investment theses, ratings, price targets and conclusions, has been independently reviewed and verified by the research analyst(s) to ensure accuracy and professional integrity.

Investors should note that ETF investments are subject to market risk, sector concentration risk, currency risk and valuation risk. Past performance is not indicative of future performance. Investors should consider whether the product is suitable for their investment objectives, financial situation and risk tolerance.