Alibaba Group Holding Limited

AI + Cloud momentum is the core positive; management sees a multi‑year step-change in TAM - Cloud Intelligence Group external revenue growth accelerated to 36% (vs 29% prior quarter); AI-related product revenue grew triple-digit YoY for the 10th consecutive quarter. - Cloud Intelligence Group’s adjusted EBITA margin held ~9% (stable despite continued investment). - Eddie explicitly frames the market shift as the move into an “AI agent era,” arguing enterprise AI budgets change: token usage becomes part of operating/R&D costs rather than only an IT services budget. - Strategic forecast/goal: over the next 5 years, Alibaba targets >$100B combined cloud + AI external revenue (including MaaS) (no CAGR quantified; management said it won’t be linear).

“ATH” (Alibaba Token Hub) re-organization is meant to tightly integrate models + applications (agents) - ATH is positioned as the organizational backbone to coordinate Alibaba’s AI stack across model (Qwen, Tongyi/MaaS foundation), applications (Qwen App for consumers; Wukong for enterprises), and infrastructure enabling layers. - Key “organizational goal”: in the agent era, Alibaba believes success requires tight integration between application and model, driven by customer-side data feedback loops (“data flywheel”). - Consumer scale: Qwen consumer-facing MAU surpassed 300 million; Qwen app integration expanded across Alibaba ecosystem apps (Taobao Instant Commerce, Alipay, Fliggy, Damai, Amap). - Enterprise scale: Wukong described as an “AI native enterprise grade agent platform,” positioned as the interface for AI capabilities compatible with enterprise data permissions/workflows. - Surprise/important detail: token distribution is described as happening primarily via the AI application layer, with MaaS as the “hub” that supports internal and external/industry use cases.

Cloud infrastructure + chips (T‑Head) are emphasized as both growth enabler and supply resilience - T‑Head achieved scaled mass production; as of Feb 2026, 470,000 chips shipped cumulatively. - >60% of ships go to external customers (public/hybrid via Alibaba Cloud), with adoption across industries (Internet finance, autonomous driving, intelligent manufacturing); chips used for both training and inference per management. - T‑Head supports Alibaba’s differentiation: cost effectiveness via co-design with Alibaba Cloud + Qwen; also supply assurance given expected global compute tightness (management explicitly called China compute supply a bottleneck over 3–5 years). - Spin-off rumor addressed cautiously: management said it’s possible but no definitive timeline for an IPO/spin-off of T‑Head (important investor uncertainty remains).