XPENG INC.(9868.HK)：NON-VEHICLE BUSINESS AND PROSPECTS COULD SUPPORT VALUATION PREMIUM BUT YET TO JUSTIFY VALUATION SWITCH TO SOTP

Core Viewpoint - XPeng's 3Q25 total revenue increased by 11.5% QoQ to RMB20.4 billion, but the revenue structure showed mixed results with vehicle sales underperforming due to a decline in average selling price (ASP) while service revenue surged, driven by technological services from Volkswagen [1][2][3] Revenue Performance - Total revenue for 3Q25 reached RMB20.4 billion, reflecting an 11.5% QoQ growth, aligning with estimates, but vehicle sales only grew by 6.9% QoQ, missing projections due to a significant drop in blended ASP [2] - Services and other revenue nearly doubled QoQ to RMB2.33 billion, exceeding expectations due to milestone payments from Volkswagen [2] Margin Analysis - Vehicle margin decreased by 0.8 percentage points QoQ to 13.1%, slightly below estimates, attributed to weaker demand for key launches and early-stage deliveries of new models [3] - Blended gross margin improved by 2.8 percentage points QoQ to 20.1%, driven by high-margin service revenue, which accounted for 42% of gross profit [3] Delivery Guidance - 4Q25 delivery guidance is set at 125,000-132,000 units, indicating a potential pullback in November deliveries due to softer end-user demand and delays in new models [4] - For 2026, XPeng plans to launch three super EREV products and four new dual-powertrain models, aiming to capture niche demand trends [4] AI and Technological Advancements - At the 2025 AI Tech Day, XPeng introduced its AI blueprint, including next-gen AI models and robotaxi capabilities, positioning itself as a leader in AI-driven mobility [5] - Plans for three Robotaxi models and mass production of humanoid robots by the end of 2026 were announced, indicating ambitious expansion into AI-driven sectors [5] Valuation Adjustments - Revenue forecasts for 2025-26 have been revised down by 6-7% to RMB77.2 billion and RMB110.3 billion, respectively, reflecting weaker delivery guidance and ASP erosion [6] - Despite recent AI advancements, it is deemed premature to shift valuation methodologies due to the early stage of these initiatives [6][7]