可靈起飛，老業務趴地，快手還能繼續走 AI 重估路嗎？

$KUAISHOU-W(01024.HK) released its Q4 report after the Hong Kong close on Mar 25 Beijing time. Headline results were slightly ahead of expectations, and the key bright spot was stable topline growth with internal efficiency gains driving margins. That said, Kling, the business underpinning much of the valuation upside, fell short of early-year bullish sell-side expectations, and rising competition likely contributed to the share pullback since Feb.

With the stock now trading at under 9x PE, we think most of the negative has been priced in after the recent correction. Management guided for Kling to at least double this year on the call, suggesting Kuaishou’s AI optionality is not fully reflected in the price.

Details as follows:

1. Operating efficiency kept improving: AI capex is flowing through (depreciation +10% YoY; R&D +34% YoY), but Kuaishou tightened opex levers, including personnel costs up only 4% and marketing flat YoY. As a result, core OPM reached 13.5%, expanding 350bps YoY and ticking up 40bps QoQ. We expect AI investment to weigh more on profits as related revenue scales. Whether the market accepts the impact on operating leverage without compressing multiples will hinge on either a re-acceleration in legacy businesses or Kling’s ability to defend against competition and industry shifts while sustaining rapid sequential growth.

2. Kling missed early-year sell-side hopes: The Dec launch of Kling 2.6, with improved motion control and audio-video co-generation, earned strong user feedback and pushed monthly gross billings to a record US$20mn. Off that, some analysts lifted Q4 revenue forecasts from guidance of RMB300mn to RMB350–400mn, and projected FY26 revenue near RMB2bn, implying nearly 2x growth. Actual Q4 revenue was RMB340mn, near the low end of guidance and market expectations. Companies typically preview with analysts, and data showed iOS momentum faded quickly in Jan after Dec’s spike; Feb’s Kling 3.0 update coincided with Seedance 2.0 going viral, capping a rebound in billings. It is likely the Street trimmed estimates accordingly, pressuring Kuaishou’s shares since Feb.

On the call, management disclosed Kling’s Jan ARR exceeded US$300mn, up 25% MoM vs. Dec, implying billings were not as weak as some external datasets suggest. We believe third-party trackers cover mainly the iOS app, while Kling’s enterprise use cases, better captured on PC and accounting for a rising mix (40% in Q3), are underrepresented in those datasets.

3. E-com growth steady: Q4 e-com GMV rose 13% YoY, slowing vs. Q3, with full-year GMV near RMB1.6tn. With creator-driven distribution, full-service operations, and super links, take rate improved by 8bps YoY (down QoQ due to peak-season rebates), lifting e-com revenue growth to ~22% YoY, a touch above top-tier house estimates (~18% YoY).

4. Ads slightly beat: Q4 online marketing revenue grew 15% YoY, continuing a modest recovery from Q3 and slightly topping expectations. Given macro consumption softness, intensified competition, and traffic tax concerns, the bar was low. Whether the traffic tax impact could broaden further into 2026 is a key point to watch from management’s commentary. Growth benefited from an easy base, resilient e-com, continued strength in short dramas (incl. AI comics) and mini-games, as well as the ad tech stack upgrade with the OneRec recommender, UAX auto-bidding (80% penetration of out-of-network ad spend, +10ppt QoQ), and incremental AIGC ad tools.

5. Live streaming under expected pressure: Q4 live-streaming paid spend fell 2% YoY, reflecting sectoral contraction and Kuaishou’s enforcement to sustain a healthier ecosystem.

6. User metrics stable: The platform is mature, so user KPIs were steady: MAUs 740mn and DAUs 410mn, both up at low single digits YoY. Stickiness was 55%, and time spent was flat YoY.

7. Shareholder returns may step up: With the Q4 share pullback, buybacks accelerated; full-year repurchases reached HKD3.2bn, and the company announced a HKD3.0bn dividend, implying a combined shareholder return of 2.7%. Kuaishou holds ample cash and shares have come under renewed pressure, so we expect sizable buybacks to continue post quiet period, subject to management guidance on the call.

8. Financial details at a glance

Dolphin Research View

Q4 was decent, especially against an implied market bar embedded in ~9x P/E; expectations have turned more conservative. Still, concerns are not unfounded: user growth is near stall speed and offers limited cushion for legacy segments amid policy headwinds (traffic tax and soft goods consumption) and tougher competition. In short, Kuaishou’s growth story leans on AI: ad-side lift, consumer-facing AI comics, enterprise AI marketing and AIGC solutions, and the pivotal Kling business.

AI on the ad side primarily slows the deceleration in revenue growth, while Kling is genuinely incremental and the part that excites capital. Since Kling’s GTV inflected in Q2 last year, investors quickly re-rated Kuaishou, with many underwriting high growth and assigning ~35x P/S to Kling. This high multiple makes short-term volatility in Kling’s billings disproportionately impactful to Kuaishou’s overall valuation. For example, ARR at US$150mn vs. US$300mn could swing valuation by roughly HKD35bn (~15% of current market cap), and last year’s longer-term revenue optimism was even higher, implying larger swings.

We see FY26 expectations as broadly neutral:

1) Legacy businesses are modeled to grow 8.5% YoY, down from 11.7% this year, baking in headwinds from the traffic tax and competition. 2) Helped by highlights from Dec’s v2.6 and Feb’s v3.0, management disclosed Kling’s Jan ARR surpassed US$300mn and guided at least 2x growth in 2026 (RMB2.1bn+), above the Street’s +85% YoY. iOS billings trends that the market follows miss the growing B-side on PC, which is harder to track but increasingly material.

Assuming a broadly neutral revenue outlook, we further haircut margins for AI spend: we model a 200bps drag from AI (RMB14bn capex, 5-year depreciation), partly offset by 150bps of cost efficiencies. Net-net, we reduce the 2025 margin by 50bps. At ~US$30bn market cap, the implied PE on adjusted legacy earnings is ~9.5x, which looks broadly fair vs. e-com peers.

In other words, the current market cap prices in little for Kling. Even if the old-school label has not fully faded, Kuaishou’s AI tag should not be ignored. Let us run the math:

On a cautious 8x PE for legacy, fair value is ~US$25bn. With management guiding Kling to at least US$300mn revenue this year, applying 30–40x P/S in line with comps implies US$9–12bn for Kling, i.e., 15–25% upside from current levels.

If we go more conservative, acknowledging lower investor appetite for single-model plays and strong competition from Seedance, a 20–30x P/S still values Kling at US$6–9bn, or ~5–15% upside.

Detailed charts below