Alibaba Group Holding Ltd - ADR

AI + Cloud is clearly accelerating and management sees an “inflection → scale” phase - Group total revenue: RMB 243.4B (+11% YoY, like-for-like excluding Sun Art & Intime). - Cloud Intelligence Group external revenue: +40% YoY; AI-related product revenue: triple-digit growth for the 11th consecutive quarter. - AI-related product revenue run-rate: annualized ~RMB 35.8B; AI products are ~30% of external cloud revenue. - Forecast/outlook: management expects AI-related product revenue to cross 50% of Cloud external revenue in ~1 year, making AI the primary growth engine; expects Cloud external growth to accelerate beyond the current 40% rate. - Why it matters: AI-related software/services are positioned as a higher-growth, higher-quality margin stream versus traditional compute/storage.

MaaS / Model Studio ARR target disclosed for the first time—strong opportunity but relies on scaling compute - Management provided their first disclosed ARR figure/trajectory: Model and application services ARR (incl. Model Studio) - >RMB 10B annualized in the June quarter - ~RMB 30B by year-end - Revenue mix on MaaS: mainly API/model usage on the Bailian (Model Studio) platform for now; platform is open to both Alibaba proprietary models (e.g., Qwen, voice/video) and third-party/open models. - Surprising / positive: despite “agentic” trends (higher inference cost), management claims customer willingness and pricing are improving (token price hikes cited) and that there is still supply scarcity—many customers are waiting to access services. - Implication for investors: if token demand keeps outpacing supply, MaaS growth could remain strong; if compute supply expands slower than demand, upside remains but margin dynamics depend on supply cost.

Infrastructure moat: T-Head GPUs and chip supply scaling are central to margin and capacity - T-Head proprietary GPU chips are in scaled “MaaS production”: >60% of compute capacity already serving external customers across Internet, financial services, autonomous driving. - Management emphasizes being the only China AI cloud provider with self-developed AI chips at scale, aiming to improve autonomy of compute supply and support gross margin uplift. - Forward-looking margin expectation: management expects material gross margin improvement as MaaS scales and as T-Head deployment ramps; specifically, they suggested better margins could start appearing in ~1–2 quarters, with higher gross margin over the next 2–3 years. - But: they also acknowledged chip/memory physical constraints and cost inflation (cost of deploying new servers cited as >2x vs a year ago, i.e., over 100%).