CSC: Hang Seng Tech Awaits AI Re-rating; US Stocks Must Guard Against Elevated Valuations Amplifying Earnings Verification Risk
nashnova research
CITIC Securities' 2027 overseas strategy calls Hang Seng Tech structurally misaligned with China's AI wave, making a broad rally unlikely; US mega-cap tech faces capex depreciation, antitrust headwinds, and concentration risk simultaneously, with AI-spend-to-profit conversion as the year's pricing throughline.
Why did Hang Seng Tech underperform globally in 2026?
Two forces converged: offshore rate hikes suppressed valuation recovery, while the index itself was loaded with the wrong cargo — internet and auto cyclicals dragged it down alongside China's macro weakness.
Semiconductors and AI software, the sectors that actually track the AI trend, account for less than 20% of the index weight.
This means → Hang Seng Tech lagged not because China's AI story is weak, but because the index composition missed the industry shift.
Is a broad Hang Seng Tech rally likely in 2027?
CITIC's call: a broad rally is unlikely; structural, selective opportunities are more probable.
On the positive side, modest nominal-growth recovery and easing price wars should lift core operating profits. On the negative side, AI capex depreciation is hitting income statements and cash flows all at once.
Three potential catalysts: rebalancing at extreme valuations, cooling rate-hike expectations, and a re-rating of internet-AI narratives. Preferred lanes: high shareholder-return names, niche e-commerce and gaming, select AI plays.
Can internet platforms turn AI spending into profit?
The key question has shifted from "can models keep up?" to "can investment convert into earnings?"
In plain terms = the tech-catch-up phase is over; the market now wants to see the books.
Benchmark estimates: Alibaba's IaaS ROIC (return on invested capital — how much each dollar of investment earns) is 13.1%, MaaS is 17.7%; Tencent's Xiaowei sits at roughly 16.4%.
AI-gateway migration does not necessarily drain Chinese platforms' commercial value — data licensing, payment rails, and offline fulfillment form ecosystem moats that are hard to bypass near-term.
What risks face US mega-cap tech?
CITIC's view: the problem is not AI demand itself, but that failure to verify returns gets amplified by high valuations and high concentration.
Three pressures hit earnings quality at once: capex second derivative turning negative, depreciation front-loading, and unresolved antitrust proceedings.
The big four cloud players are diverging: Amazon has the strongest cash-conversion path (AI workloads lift CPU cloud demand + high-margin advertising); Microsoft leans on Copilot's enterprise penetration, delivering steady B2B monetization; Alphabet is expanding search and cloud in tandem, but quarterly free cash flow has turned negative — conversion of its $707 billion procurement commitment into revenue is the swing factor; Meta is improving ad efficiency, yet capex at 51% of revenue plus litigation costs leave the thinnest free-cash-flow cushion.
How should investors pick software names?
The report urges a shift from concept trades to a "three-haves" framework: cross-system workflow data, high net revenue retention (NRR — the rate at which existing customers renew and expand), and high free-cash-flow margins.
Top pick: ServiceNow — it controls the cross-system action layer, with 50% of new business already migrated to non-seat-based billing. SAP benefits from deep data-layer moats and a 2027 cloud-migration cycle.
Salesforce is rated underweight: organic growth is slowing and its interface layer is being ceded. In data infrastructure, Snowflake leads (NRR at 126%); MongoDB serves as a high-beta trading vehicle.
What else matters — consumer internet and autonomous driving?
E-commerce: overall demand remains soft, but improving competitive structures are now the dominant variable — rising take rates matter more than GMV growth. Cross-border e-commerce faces US-EU de minimis repeal and ESG legislation, systematically narrowing its direct-mail cost edge.
Niche verticals (pharma, MRO — industrial procurement, second-hand consumer electronics) are among the report's top picks: low penetration means growth is driven by category expansion, and profit growth systematically outpaces revenue growth.
Meituan: CLC (community group-buy and local commerce) has turned profitable; food-delivery unit economics are positive. Yet Keeta may be the biggest expectations gap — the market expects harvest mode, while a two-front campaign in Brazil and potential Southeast Asia expansion could mean more spending.
Autonomous driving: L2 enters tighter regulation, L3 moves from pilots to limited commercialization, L4 / Robotaxi enters scaled verification with US-China momentum. This reflects a broader industry shift from narrative to proof — mirroring the same verification logic facing US mega-cap tech.
市场有风险，内容仅供研究参考，不构成投资建议。