Alibaba’s earnings report this week reads as two companies stapled together. One is dying slowly. The other is eating everything in its path.
The headline numbers look contradictory. Net income collapsed 76% year-over-year to 10.54 billion yuan ($1.55 billion) for the quarter ended June 30. Free cash flow ran negative by more than $6.6 billion. Capital expenditure jumped 75% to 67.68 billion yuan ($9.98 billion), and nearly all of it went into chips, data centers, and compute capacity. By any traditional read, that’s a company bleeding money.
Revenue tells a different story. It climbed 9% to 268.95 billion yuan ($39.64 billion), edging past the roughly 268.9 billion yuan analysts expected, according to Alibaba’s official quarterly filing. Buried inside that top line sits the number that actually matters: AI Cloud and Compute Services grew 45% year-over-year to 48.44 billion yuan ($7.14 billion), its fastest pace in 22 quarters. Segment adjusted EBITA more than doubled, up 133% to 5.63 billion yuan.
This isn’t a company losing control of its spending. It’s a company shifting capital from a mature, low-growth retail business into a young, high-growth AI infrastructure business, and doing the math in public.
The real story isn’t the miss. It’s the streak.
Most coverage will lead with the profit drop, because that’s the number that moves a stock price on a Thursday morning. Shares fell around 4% in premarket trading, Reuters reported. Fair enough.
But sit with this number instead: AI-related product revenue inside the cloud segment hit 12.38 billion yuan. That’s the twelfth consecutive quarter of triple-digit year-over-year growth for that line item. Three years of compounding demand for Alibaba’s AI products, uninterrupted, through a stretch when China’s broader consumer economy stayed visibly weak.
That streak is the real signal. Quarterly profit swings with capex timing and chip pricing. A three-year growth streak in AI product adoption tells you something about durable enterprise demand — the kind that doesn’t vanish because a data center bill came in high one quarter.
Why the losses are the point, not the problem
Compare Alibaba’s posture to Microsoft, Google, and Amazon. All three have poured record capex into AI infrastructure over the past two years while cloud margins temporarily compressed. Markets rewarded all three anyway, because they understood the pattern: infrastructure spend now, monetization later, at scale.
Alibaba isn’t starting from zero here. Its cloud unit already ranks as the world’s fourth-largest infrastructure-as-a-service provider and the largest in Asia Pacific by revenue, per Gartner’s April 2026 rankings. It holds roughly 37% of China’s cloud infrastructure market, well ahead of Huawei’s 17% and Tencent’s 10%, according to Omdia data cited ahead of this earnings print. Alibaba isn’t building AI infrastructure from scratch. It’s reinforcing a lead — and that lead still runs into the same hard constraint every AI buildout hits eventually: enough chips and enough power to actually run them, a bottleneck showing up in how far infrastructure operators are already going to source both.
Meanwhile, the segment posting the steepest losses tells a messier story. AI Labs and Applications — home to the Qwen consumer app and the QwenWork enterprise agent — widened its adjusted EBITA loss to 13.86 billion yuan, up from 3.22 billion yuan a year earlier. That’s not a red flag on its own. Every hyperscaler currently eats losses on consumer-facing AI products while inference costs stay high and user bases scale. The real question isn’t whether Qwen loses money today. It’s whether Alibaba can turn that user base into the kind of recurring enterprise revenue Cloud already generates.
What this earnings call actually tests
CEO Eddie Wu calls it a “full-stack AI strategy”: chips through T-Head, infrastructure through Cloud Intelligence, models through Qwen, all under one roof. That vertical integration is the bet. Control enough of the stack, and AI demand growth compounds across every layer instead of leaking margin to a partner at each handoff.
The uncomfortable part for investors: this bet requires exactly the kind of quarter Alibaba just posted. Heavy spend. Thin near-term profit. A stock that dips on the headline number. The comfortable part: the underlying growth curve, twelve quarters running, hasn’t bent yet, a point South China Morning Post’s earnings coverage also flagged as the quarter’s standout.
So which is it? Has AI infrastructure demand from Chinese enterprises hit its ceiling, or is Alibaba still early in a multi-year build-out? This report doesn’t answer that definitively. But a twelve-quarter streak makes a far stronger case for “early” than any single quarter’s profit number ever could.
Related: China Is Giving Away AI. The Real Product Is Governance
