Kingsoft Cloud Holdings Limited (KC) Q4 2025 Earnings Call Transcript

Kingsoft Cloud hosted its Q4 and full-year 2025 earnings conference call on March 25, 2026; the associated earnings release was distributed and is available on the company's IR site. Management (Chairman/Acting CEO Zou Tao and CFO Li Yi) reviewed business strategy and financial performance, with a Q&A including analysts from CICC, CLSA and Goldman Sachs. The provided excerpt contains no financial metrics, guidance, or material operational updates.

Analysis

Kingsoft Cloud’s optionality is now more about mix-shift than pure scale: winning vertical enterprise AI workloads materially raises ARPU and gross margins because managed AI services retain customers longer and carry higher implementation fees. That creates a two-stage cash profile — near-term capex and margin volatility as GPU/accelerator purchases spike, and a multi-year FCF uplift once software + HA-managed services scale to 20–30% of revenue.

Competitively, mid-tier clouds can win through vertical productization and custom SLAs where hyperscalers compete on scale but not depth; success here produces stickiness that is hard for large, generalized players to pry off without discounting. A second-order supply-chain effect: aggregated GPU demand from second-tier clouds can re-route procurement flows toward local silicon partners and EMS providers, pressuring global OEM lead times and creating arbitrage opportunities for alternative accelerator vendors.

Key catalysts and horizons: near-term (days–weeks) price action will be driven by guidance credibility and data-point releases on large AI customer wins; medium-term (3–12 months) earnings cadence and capex timing will reveal whether GPU purchases compress margins or simply pre-fund higher ARPU; long-term (1–3 years) upside hinges on software monetization and any changes to export/control regimes that affect access to premium accelerators. Tail risks include aggressive price competition, a macro IT budget swoon, or export constraints that impair access to high-end accelerators and reverse the upside.

From a risk-management standpoint, treat any equity exposure as a convex bet on AI monetization, not base-case infrastructure scale. Use staged entries around verifiable customer wins and capex disclosures, and prefer option structures that limit downside while keeping asymmetric upside to a 12–24 month re-rating.

Key Decisions for Investors

• Long KC equity (KC) — entry: accumulate on a >3–5% post-earnings intraday dip or on report of a named AI enterprise win. Position size: 2–3% of portfolio. Target: +40–60% in 12 months if managed AI revenue reaches ~20% of sales; Stop-loss: -25% from entry. Rationale: convex optionality from software/managed services monetization while limiting near-term capex draw risk.
• Pair trade: long KC / short BABA (1:1) — timeframe 6–12 months. Size: net market-neutral dollar exposure, 1–2% gross each. Target: capture 25–35% relative outperformance if KC wins vertical share; Stop: cut if spread widens 15%. Rationale: expresses share-shift to a vertical-focused mid-tier cloud versus a broad hyperscaler vulnerable to commoditization.
• Options collar to buy convexity — buy KC Jan-2028 LEAPS calls (near-ATM) financed by selling KC Oct-2026 OTM puts. Capital: ~1% portfolio net debit. Risk/Reward: defined downside (puts assigned) with multi-bagger upside if AI monetization accelerates. Use this to get long-dated optionality while reducing cash outlay.
• Short-duration hedge on GPU supply risk — buy options on large GPU supplier(s) (e.g., NVDA calls) or an AI infra ETF to pair with KC exposure over 3–12 months. Notional: 0.5–1% of portfolio. Rationale: captures upside if GPU scarcity and price inflation drive higher cloud ASPs that benefit KC’s higher-margin managed offerings.