Alibaba disclosed its latest Form 13F with the SEC this week, revealing $653 million in US holdings split between XPeng (88%) and Weibo (12%). The wire services treated it as news. It's background noise next to what the company actually did three days earlier.
On May 13, Alibaba reported Q4 FY2026 earnings. Revenue came in at RMB 243.4 billion, up 3% year-over-year, missing consensus by a hair. EPS was $0.61, down sharply from the prior year. The stock rallied anyway. Wall Street's take: "mixed results," "earnings evaporated." The headline narrative focused on the miss.
The narrative is wrong. The story isn't the miss. It's where the money went.
Cloud Intelligence grew 38% - and AI is the engine. Revenue from Alibaba's Cloud Intelligence Group jumped 38% year-over-year to RMB 41.6 billion ($6.13 billion), and external cloud revenue accelerated even faster at roughly 40%. AI products now touch 30% of all external cloud customers, up from single-digit penetration a year ago. This is the inflection point. When AI workloads reach that share of the customer base, it means the platform has moved past early adopters and into mainstream adoption - the point where growth compounds rather than plateaus. Management said AI spending will exceed original targets. That's not a warning. It's a commitment.
The operating loss is a choice, not a collapse. Alibaba posted a Q4 operating loss as it poured capital into AI infrastructure, model development, and data center capacity. Management called it an investment. That matters because it means the earnings miss is discretionary - the company chose to reinvest rather than protect the bottom line. There's a difference between a business that can't earn and one that won't, at least not yet. The former is a dying moat. The latter is a build phase.
The 13F filing is irrelevant to the thesis. Alibaba's US portfolio - $653 million across two names - is a rounding error against a company generating RMB 243 billion in quarterly revenue. Yes, the XPeng bet shows management is betting on China's EV story alongside its own. That's a separate conversation. The 13F doesn't tell you anything you need to know about Alibaba's earnings power.
The valuation disconnect is already forming. At roughly $141 per share, BABA trades at about 20x forward EPS. By FY2028, that multiple compresses to around 15x on consensus estimates. That's the GARP number. The question is whether Cloud Intelligence continues accelerating and whether the AI investment cycle converts into margin expansion. If cloud revenue keeps growing in the 30-40% range and AI penetration moves from 30% to 50% of customers, those FY2028 estimates get revised up - which means the stock re-rates even at the same multiple. The math works if the execution holds.
The key risk is China. A deeper-than-expected economic slowdown could drag on the core commerce business, which still generates the majority of revenue. If e-commerce growth stalls and cloud can't offset it, the AI investment thesis becomes a drag rather than a catalyst. That's the condition that breaks the setup.
At only 15x FY2028 EPS targets, the stock doesn't price in what happens if Alibaba's cloud growth stays above 30% while AI product revenue scales. The 13F is yesterday's headline. The cloud numbers are the story.