Alibaba Group Holding Ltd - ADR

Cloud + AI momentum is the clear growth engine (and accelerating): - Alibaba Cloud external revenue grew 45% YoY in the June quarter (described as a 22-quarter high). - AI-related product revenue grew triple-digits for the 12th consecutive quarter, reaching RMB 12.4bn (≈ RMB 49.5bn annualized, ~$7.3bn). - AI-related products = 35% of external cloud revenue (rising share supports mix/monetization strength). - Management expects cloud growth to remain on an accelerated trajectory in coming quarters; they also guided an annualized AI-quarter revenue approaching ~$10bn in the next quarter (implying further step-up vs current run rate).

Profitability improving, but overall earnings/cash flow were pressured by infrastructure spend (good long-term, bad short-term): - Cloud adjusted EBITDA margin reached 11.6% this quarter (Toby also cites 12% for AI-related product-driven cloud profitability); management expects further steady margin expansion sequentially. - Group adjusted EBITDA fell 30% YoY to RMB 27.3bn, primarily due to technology/infrastructure investment, partially offset by cloud operating improvements. - GAAP net income fell 75% YoY to RMB 10.4bn (management attributes to lower operating income and investment-related items like mark-to-market/disposals). - CapEx was RMB 67.7bn (up sharply YoY); management clarified the spike was partly due to equipment delivery/ procurement cycle volatility and higher CPU procurement driven by agent demand + component price increases. - Free cash flow was an outflow of RMB 44.7bn (worse vs prior year outflow), mainly from AI infrastructure investment.

Full-stack AI commercialization is the strategy (compute + MaaS + proprietary silicon), with supply constraints turning into an opportunity: - Alibaba is upgrading toward an “agentic cloud.” Management said MaaS ARR exceeded RMB 16bn (as of August) and frames this as evidence of accelerating commercialization. - Proprietary chips are scaling: T-Head chips serve >650 customers on Alibaba Cloud; Zhenwu M890 supernode launched at commercial scale. - They claim hyperscale AI data center delivery time has been cut to ~100 days (to speed infrastructure build-out). - Management expects compute demand to outstrip supply in the near term; as they ramp supply, AI/cloud revenue growth should accelerate further and profitability improve.