Kingsoft shares jump as FY profit climbs on investment gains, AI push

Profit attributable rose 29% y/y to ¥2.00bn (EPS ¥1.46 vs ¥1.16), while shares jumped as much as 9.4% to HK$25.10. Total revenue fell 6% to ¥9.68bn and operating profit dropped 51% to ¥1.78bn as R&D and marketing ramped up; net other income included a ¥1.25bn one-off gain from deemed disposal of Kingsoft Cloud shares. The company proposed a final dividend of HK$0.13 (down from HK$0.15) and reiterated plans to invest in AI-driven office tools, expand internationally, and develop new game titles.

Analysis

Earnings quality appears bifurcated: operating momentum is soft while headline profitability is propped by non-recurring items and aggressive capital allocation toward R&D/marketing. That mix creates a volatile baseline — markets should treat current margins as provisional until recurring monetization from AI-office products and new game titles prove consistent over multiple quarters.

Strategically, the company is pivoting toward enterprise AI workflows which has two second-order effects: it raises lifetime customer acquisition costs (enterprise sales cycles and channel investment) and increases near-term demand for compute and storage equipment. Suppliers to the AI stack (servers, GPUs, systems integrators) and regional cloud partners stand to capture upside if the pivot scales; conversely, legacy gaming peers with weaker pipelines face secular pressure on monetization and user retention.

Key risks are concentrated and time-staggered — regulatory and macro shocks can truncate both product launches and enterprise spending within 0–3 months, while successful monetization or a meaningful stake monetization/spin would likely materialize over 3–12 months. The single biggest reversal trigger is failure to convert AI engagement into subscription or usage revenue at acceptable gross margins, which would force either deeper cuts to capital returns or heavier dilution.

Contrarian read: the market is likely oscillating between two mistakes — overstating the one-offs as durable earnings or dismissing the AI pivot as too distant. A disciplined view is to treat the name as a conditional growth binary (execution-dependent) rather than a straight value or growth story; that framing changes position sizing and time horizon for any trade.

Key Decisions for Investors

• KC — Buy a size-limited long (e.g., 4–6% portfolio weight) with a 6–12 month horizon. Scale in on any >10% pullback; target 30–40% upside if AI office monetization starts showing sequential ARPU growth, stop-loss at -20% to cap downside if both gaming and enterprise traction disappoint.
• Pair trade — Long KC / Short APP (equal notional) for 3–6 months to express conviction in enterprise AI recovery vs. mobile-ad/gaming ad headwinds. This neutralizes macro beta; profit if KC monetizes AI while APP suffers from cyclical ad reallocation. Close if divergence <5% after 3 months.
• SMCI — Tactical long (3–6 month) to capture infrastructure restocking from increased AI R&D spend across software incumbents. Small-sized position; target +25% with trailing 15% stop given high volatility in infra names.
• Options hedge — If holding KC equity, buy 9–12 month protective puts or purchase a cheap long-dated call spread to retain upside while limiting capital at risk; use strikes that reflect a 20–25% downside protection and cap premium to <2% of position value.